sitemap.yaml

assessment.md:
  hash: 7bed94a7acfc39e185de97edacc91f93
  summary: This document outlines a collaboration opportunity for a company aiming
    to enhance its product through research and development on memory and retrieval
    features. Leveraging experience in preprocessing and retrieval, the collaboration
    seeks to accelerate R&D efforts through conversations, brainstorming, and code
    contributions. Key objectives include integrating new features into the retrieval
    engine and improving performance on critical customer queries. Various collaboration
    models are offered, including asynchronous support and involvement in technical
    writing and strategy, with corresponding fees. Success is measured by enhanced
    feature integration and customer satisfaction, with a flexible and responsive
    communication strategy.
books.md:
  hash: a490c4c020cd2048a38a2119bb4ae9d2
  summary: The list outlines a diverse range of books that have influenced the author's
    life, categorized into sections based on impact, reading phases, and genres. The
    "Influences on my life" section includes profound works like the "Tao Te Ching"
    by Lao Tzu and "Book of Disquiet" by Fernando Pessoa, highlighting personal growth
    and philosophical insights. The "Prince Era" reflects an upcoming period of reading
    European philosophy, with plans to explore existential works by Sartre and Camus.
    The "Highlights" section merges influential books across genres, featuring authors
    like bell hooks and Byung Chul Han. The "Business Stuff" section covers practical
    and strategic insights with books by Ray Dalio and Alex Hormozi, while unread
    recommendations include classics like "Man's Search for Meaning" by Viktor Frankl.
    Key themes include personal development, philosophy, existentialism, strategic
    business principles, and literary exploration.
contact.md:
  hash: 9539445a763fd86732ec6e40a220df0b
  summary: This professional offers expertise in applied AI, focusing on collaboration
    with fast-growing startups ready to invest in this field. Potential clients can
    explore detailed services on the [services](services.md) page or initiate a project
    discussion by completing the [contact form](https://form.typeform.com/to/gqgTx3I6).
    For immediate advice, a [paid call](https://cal.com/jasonliu/expert-call) is available.
    Additionally, the expert is open to speaking engagements and can be contacted
    via [email](mailto:work@jxnl.co) for events or podcasts. Key terms include applied
    AI, startups, consulting services, and speaking engagements.
expert-calls.md:
  hash: 8118860d0cc6f56abca4eb9e3fe9d163
  summary: 'Jason Liu offers expert AI consultations tailored to address your challenges,
    with two main options: Quick and Comprehensive Consultations. The Quick Consultation
    (25 minutes) is designed for focused problem-solving, while the Comprehensive
    Consultation (55 minutes) includes pre-call preparation, in-depth analysis, and
    a post-call summary memo with recommendations. These services aim to provide technical
    insights, strategy checks, and effective planning for AI initiatives, making them
    ideal for advancing projects and resolving roadblocks. Key services include AI
    strategy development, technical query resolution, and AI project planning.'
experts-and-vendors.md:
  hash: 15ee6b344563c9f8d2281e035849fcf7
  summary: Discover a comprehensive network of AI and ML specialists, vendors, and
    industry experts, offering access to top talent for strategic consulting purposes.
    Key experts include Hamel Husain, an LLM consultant; Eugene Yan, an ML Research
    Scientist at Amazon; and Shreya Shankar, a PhD student at Berkeley focused on
    practical ML solutions. Trusted vendors like Braintrust, LanceDB, and Vespa.ai
    provide cutting-edge platforms and services in LLM evaluations, multimodal AI
    databases, and search and recommendation systems. Optimize your AI and ML strategies
    with insights from leaders in AI-driven document processing, serverless platforms,
    and recommendation technologies.
index.md:
  hash: 4cd3df2d4a41d844676ca5df6b996a44
  summary: 'This website is managed by a machine learning engineer, angel investor,
    and startup advisor who specializes in consulting, open source, and applying large
    language models (LLMs). The site offers free resources for enhancing skills in
    Retrieval-Augmented Generation (RAG), AI engineering job listings, and insights
    on consulting practices. It features a comprehensive RAG course starting in February
    2024, focusing on building robust systems and optimizing AI processes. Stay updated
    with content and resources shared on various platforms like YouTube, Twitter,
    and GitHub. Keywords: machine learning, angel investor, startup advisor, RAG,
    LLMs, AI engineering, consulting, open source, course, evaluation pipelines, user
    interfaces.'
investments.md:
  hash: 02a1c61326bb80e4b068440605d1454e
  summary: 'Discover innovative companies revolutionizing their industries. Augmental
    enhances digital accessibility with its MouthPad, a hands-free trackpad operated
    by the tongue. Haize Labs develops AI tools for stress testing of large language
    models, focusing on risk identification. Exa offers an AI-powered search engine
    API for real-time web data integration, ideal for developers. Shaped AI provides
    configurable AI recommendations and search systems to boost engagement and revenue.
    Pydantic Logire delivers cloud-based inventory management solutions for seamless
    supply chain integration. Vantager transforms financial data into actionable insights
    with its AI-native investment platform. Empower Sleep leverages data-driven sleep
    care to offer personalized treatment plans and improve health outcomes. Keywords:
    wearable technology, AI-based tools, semantic search, configurable recommendations,
    inventory management, data-driven investment, sleep care.'
jobs.md:
  hash: d41d8cd98f00b204e9800998ecf8427e
  summary: Of course, I'd be happy to help with an SEO summary! However, I'll need
    you to provide the specific content you want summarized. Once you do, I'll be
    able to create a concise summary highlighting the core ideas, objectives, important
    details, and relevant keywords.
services.md:
  hash: 7d876fcb84fd4187bdaddca154529cec
  summary: Applied AI Consulting Services offers expert guidance for engineering and
    product leaders in developing and optimizing AI systems, including search and
    recommendation engines. The consultant brings over 8 years of experience from
    companies like Stitchfix, Meta, and NYU, specializing in AI integration, strategic
    planning, and team building. Key services include subscription-based support for
    ongoing assistance and project-based consulting for strategic and comprehensive
    AI solutions, with a focus on enhancing data feedback loops, overcoming integration
    challenges, and scaling AI initiatives. The consultant has worked with high-profile
    clients like Zapier and Timescale, and offers access to a network of AI and ML
    experts. Key terms include AI consulting, RAG systems, data feedback loops, AI
    integration, and scalable growth.
systematically-improve-your-rag.md:
  hash: 4e35db2028c3b104606dd7617ed678ed
  summary: Enhance your Retrieval-Augmented Generation (RAG) systems with a comprehensive
    course starting February 2024. This advanced program is designed for professionals
    aiming to build real-world RAG systems and has attracted participants from companies
    like Amazon, Adobe, and Zapier. The course offers practical, hands-on learning
    experiences, real-world case studies, and 12 hours of expert instruction, all
    aimed at improving your system's performance. Enroll to gain actionable insights,
    learn best practices, and achieve a competitive edge in AI. Additional expert
    consulting services are available for personalized AI solution development and
    strategic planning. Key offerings include RAG system enhancements, expert-led
    instruction, practical applications, and AI consulting.
user_manual.md:
  hash: 03357f7dd195f5a7f0b81618c8ef5c64
  summary: 'The content outlines an individual''s approach to management and communication,
    emphasizing the values of constant learning and growth, moving fast, and time
    freedom through asynchronous communication. Key preferences include direct and
    brief messaging, minimizing typing due to a hand injury, and prioritizing health.
    The individual prefers clear, quick-skimming messages, using Slack or text, with
    a focus on client work and delegation. Frequent updates and feedback are crucial
    to reduce anxiety and promote growth. The leadership style is hands-off, encouraging
    ownership and decision-making in the team. Keywords: management approach, asynchronous
    communication, growth, feedback, delegation, leadership style, team empowerment,
    direct communication.'
writing/index.md:
  hash: 089c8c9334b6450bf61e70922ab47876
  summary: This blog focuses on a variety of topics including consulting, open source
    projects, personal experiences, and the application of large language models (LLMs).
    It features personal writings such as advice for young people and experiences
    like adapting to losing physical abilities. The site provides a rich collection
    of short technical posts on Retrieval-Augmented Generation (RAG), exploring its
    complexities and offering insights on how to build and enhance RAG systems effectively.
    Additionally, it covers topics in technical coaching, discussing consulting strategies,
    MVP development for AI applications, and common engineering mistakes. The blog
    also links to talks and podcasts, highlighting discussions on Pydantic, AI application
    development, and data-focused platforms. Core keywords include consulting, open
    source, LLMs, RAG, technical coaching, MVP, and engineering.
writing/posts/advice.md:
  hash: 5811b313eb2cb5bf8e464e8ca50e0479
  summary: 'The blog post "Advice to Young People, The Lies I Tell Myself" by jxnl
    offers personal insights and advice for navigating life, focusing on themes like
    choice, luck, and authenticity. Jxnl emphasizes the importance of making conscious
    decisions, recognizing opportunities, and living authentically. The author shares
    personal experiences related to job searching, confidence building, and skill
    development, advocating for action despite fear and embracing simplicity over
    complexity. Other key concepts include self-worth, social media, and the power
    of writing and self-talk. The post underscores the importance of understanding
    oneself and contributing to personal growth and resilience. Keywords: life advice,
    personal development, authenticity, decision-making, confidence, self-worth, resilience.'
writing/posts/aisummit-2023.md:
  hash: 7c1ae68949492baa7248f47644d72bca
  summary: In a recent keynote at the inaugural AI Engineer Summit, speaker jxnl discussed
    the pivotal role of Pydantic in effective prompt engineering. The talk emphasized
    how Pydantic, a powerful data validation and settings management tool in Python,
    can streamline and enhance AI development processes. The presentation aimed to
    provide insights beyond traditional documentation and blog posts, offering practical
    applications for AI engineers. Viewers are encouraged to watch the full talk on
    YouTube and provide feedback to help refine future presentations. Key topics include
    AI engineering, Pydantic, and prompt engineering.
writing/posts/asking.md:
  hash: 0eaf5f876aa50f52f62be8dcb56e3acb
  summary: "The article \"How to Ask for Referrals (Among other things)\" by jxnl\
    \ offers a framework for effectively requesting referrals and other favors by\
    \ reducing the cognitive load on others. It emphasizes asking targeted, high-agency\
    \ questions\u2014such as using specific names from the person\u2019s network\u2014\
    and providing simple yes or no alternatives. Key strategies include doing preliminary\
    \ work yourself, making interactions straightforward, and avoiding generic questions\
    \ that require others to think deeply or perform additional tasks. The piece critiques\
    \ common low-agency inquiries, such as \"How can I help?\" or \"When are you free?\"\
    , and advocates for offering precise, actionable assistance instead. The focus\
    \ is on efficiency, minimizing email exchanges, and enhancing productivity while\
    \ maintaining genuine connections."
writing/posts/baths.md:
  hash: 1bb5b800fc21599774f41062562b081d
  summary: 'The article "Public Baths" contrasts the author''s experiences at bathhouses
    in Japan and the United States. In Japan, visiting onsens is described as a natural
    and spiritual experience where hygiene rituals are practiced collectively, fostering
    relaxation and connection with nature. In contrast, American bathhouses, particularly
    in LA and New York, are depicted as places focused on personal optimization rather
    than relaxation, often feeling artificial and work-like. The author highlights
    the cultural differences, criticizing American bathhouse practices as rushed and
    unsanitary compared to the tranquil, communal atmosphere of Japanese onsens. Keywords:
    public baths, American bathhouses, Japanese onsens, cultural differences, relaxation,
    hygiene, natural experience.'
writing/posts/chasing-chase.md:
  hash: c84cad30533eb77b88c7dc54a75ffcbc
  summary: 'In "Chasing Chase: A Yuppie Nightmare," the author recounts their harrowing
    experience with Chase Bank, which froze $180,000 of their funds without warning,
    leaving their small business in turmoil. Despite a decade-long loyal relationship
    with Chase, the author faced a maze of bureaucracy, lack of communication, and
    unexplained accusations, resulting in months of financial strain. This ordeal
    underscores the importance of diversifying bank partnerships, as the author ultimately
    switched to Mercury Bank, which provided efficient and supportive banking services
    for their business needs. Key points include the challenges faced by entrepreneurs,
    the need for diversified banking, and the benefits of alternative financial institutions
    like Mercury Bank.'
writing/posts/consulting-everything-i-know.md:
  hash: 50afaa899244c59e001e330b3aeb507f
  summary: 'In "Everything I Learned from AI Consulting," the author shares valuable
    insights on transitioning from a contractor to a consultant by focusing on outcomes
    rather than tasks. Key points include the importance of understanding client pain
    points, creating content to demonstrate expertise, and structuring proposals with
    multiple pricing options to reflect value rather than time. The article emphasizes
    building a sustainable consulting practice by shifting mindset, delivering high-value
    solutions, and exploring scalable products like courses and tools. Keywords: AI
    consulting, consultant vs contractor, value-based pricing, client pain points,
    proposal structure, consulting outcomes.'
writing/posts/consulting-indie.md:
  hash: 1a00796a6b2a763bc25e97231ce82dc1
  summary: 'This article explores the concept of indie consulting, contrasting it
    with traditional large-scale consulting firms like McKinsey. It emphasizes the
    value of specialized expertise, direct accountability, and long-term value creation.
    The author, faced with a career-altering injury, shifted to consulting for its
    flexibility. Key strategies include understanding the client''s value equation,
    focusing on niche markets, and leveraging content creation for marketing. The
    post also discusses the importance of establishing clear metrics, demonstrating
    expertise through situational assessments, and pricing services based on perceived
    value. Keywords: indie consulting, value creation, niche markets, content marketing,
    situational assessment, pricing strategies.'
writing/posts/consulting-listing-prices.md:
  hash: c5ee62c1f3c5eaf5574fcfdd454159f8
  summary: This article explores the drawbacks of publicly listing consulting prices
    and offers strategic pricing alternatives for consultants. Key points include
    avoiding fixed rates to prevent limiting earning potential and deterring potential
    clients, using tiered pricing models similar to airlines, and emphasizing value-based
    pricing, where prices reflect the value delivered rather than time spent. The
    piece also highlights the importance of providing minimum engagement estimates
    and using discovery calls to tailor pricing to client needs. By adopting these
    strategies, consultants can effectively price their services, attract the right
    clients, and maximize their value and earnings. Keywords include consulting pricing
    strategy, value-based pricing, tiered pricing model, and maximizing consulting
    revenue.
writing/posts/consulting-marketing.md:
  hash: 15a62226a74b9ca511432fea010da6fa
  summary: This article explores the transition from a technical role to a successful
    AI consulting business, emphasizing key strategies in marketing, positioning,
    and scaling. The core strategies include generating leads through content creation
    on platforms like Twitter and blogs, leveraging data flywheels, and utilizing
    frameworks like the "One Page Marketing Plan" and situational assessments. The
    author shares insights on how creating valuable content and structuring client
    proposals have driven significant growth, highlighting the importance of strategic
    marketing and thought leadership in the AI/ML space. Keywords include AI consulting,
    marketing strategies, lead generation, content creation, and situational assessments.
writing/posts/consulting-predictions.md:
  hash: 4ed4aa4d6599c76abb61ae374a0631e9
  summary: 'The "Why Freelancers Will Win the AI Gold Rush" article discusses how
    AI freelancers and consultants are poised to capitalize on the burgeoning AI Boom,
    similar to web developers in the 1990s and mobile developers in the 2000s. It
    highlights the unique demand for AI specialists due to the complexity of building
    AI systems and the widespread need across industries. Freelancers are positioned
    as "architects" who can leverage their skills to work on various AI projects,
    enhance productivity, and gain a compounding advantage over time. The article
    emphasizes the limited window of opportunity in AI freelancing before the market
    becomes saturated, urging those with technical expertise to swiftly engage in
    freelancing and build their reputation. Keywords: AI freelancers, AI specialists,
    AI boom, freelancing, consulting, technological revolution, AI systems, AI architects.'
writing/posts/consulting-referrals.md:
  hash: f46df4e82113f4657d73e27bffe44910
  summary: 'The article "Affective Consulting: Referrals" offers a comprehensive guide
    on mastering the art of asking for and handling referrals in the consulting industry.
    It emphasizes the importance of timing when requesting referrals and provides
    a structured approach to engage with clients effectively. Key strategies include
    getting clients to articulate the value they received, asking for specific referrals
    related to problems you''ve solved, and guiding the introduction process to ensure
    a warm and productive connection. The article also highlights the importance of
    respectful follow-ups and focusing on potential challenges rather than solutions.
    By implementing these techniques, consultants can build trust, avoid resistance,
    and position themselves as effective problem-solvers. Keywords: consulting, referrals,
    client engagement, problem-solving, business growth strategies.'
writing/posts/consulting-stack.md:
  hash: 9068d7ae2cc9745443f11c694fcea33f
  summary: This blog post provides a comprehensive guide for indie consultants looking
    to optimize their business operations using a curated "consulting stack" of essential
    tools and services. Key components include Doola for LLC setup, Mercury for business
    banking, Stripe for invoicing, and Dropbox Sign for contract management. Additionally,
    tools like Cal.com for scheduling, Circleback for meeting notes, and Anthropic
    Claude for AI-assisted content creation are highlighted. The post emphasizes the
    importance of establishing a strong business foundation, managing finances, and
    effectively scaling operations while offering practical checklists for implementing
    these tools. Key topics include tax benefits, content strategy, hiring an EA,
    and standardizing engagement processes.
writing/posts/couch.md:
  hash: 4c497c1257bb7eb7b10bc5fd6e3532f8
  summary: The article presents a critique of couches, arguing that they limit the
    functional and aesthetic potential of living spaces by imposing a fixed orientation
    typically towards a TV, promoting passive consumption rather than active engagement.
    It explores how this arrangement affects social interactions, creating awkward
    dynamics and enforcing unnecessary physical distances. The author suggests alternatives,
    such as using modular seating options like rugs or bean bags, and drawing inspiration
    from multipurpose spaces like tatami rooms, which offer the flexibility to support
    various activities and enhance personal interaction. The critique also emphasizes
    the health implications of prolonged sitting, advocating for more dynamic seating
    arrangements to foster movement, improve health, and maximize space utilization.
writing/posts/data-flywheel.md:
  hash: eb73c54554ac3fe15112aeab5080c64f
  summary: 'The article "Data Flywheel Go Brrr: Using Your Users to Build Better Products"
    explores how businesses can leverage user interactions to enhance product development
    through data flywheels. It emphasizes the importance of collecting user data to
    create self-reinforcing systems that continually refine AI models, enhancing product
    quality and user engagement. Key examples include Netflix and Spotify, which use
    data to personalize content and attract more users, creating a virtuous cycle.
    The article highlights the need for aligning data flywheels with business objectives
    to drive tangible results, urging companies to embrace this strategy in the AI-driven
    future. Key topics include network effects, data collection, AI improvement, and
    user-centric product design.'
writing/posts/data-literacy.md:
  hash: a621f249b559728ba5ca062ca62c6e1b
  summary: 'The article "10 Ways to Be Data Illiterate (and How to Avoid Them)" outlines
    essential strategies for AI engineers to improve data literacy, ensuring accuracy
    and impact in data-driven projects. It addresses common pitfalls such as ignoring
    data quality, lack of domain understanding, improper data splits, and overfitting
    models, and provides actionable solutions like conducting exploratory data analysis,
    engaging with domain experts, and using appropriate evaluation metrics. Key topics
    include data visualization, understanding sampling bias, ensuring data privacy
    and security, and distinguishing between correlation and causation. Emphasizing
    interpretability and explainability is also highlighted as crucial for building
    transparent and trustworthy AI systems. Keywords: data literacy, AI engineers,
    data quality, domain understanding, model evaluation, interpretability.'
writing/posts/freediving.md:
  hash: 3fb6111a3a3fdeb249dbe17eea207361
  summary: The article "Freediving under Ice" shares a personal journey of discovering
    freediving as a passion, particularly after suffering hand injuries during the
    COVID pandemic that restricted other physical activities. Initially engaging in
    various sports like ping pong, rock climbing, and jiu-jitsu, the author shifted
    focus to swimming and freediving, culminating in a thrilling ice dive experience
    in Northern Ontario, Canada. This adventure sparked a deep interest in the sport,
    with future aspirations to explore underwater environments and wildlife across
    the world. Key topics include freediving, ice diving, physical fitness, and personal
    growth through overcoming physical challenges.
writing/posts/getting-goals.md:
  hash: 3661dd1fb07440c37771f3d73f04c13f
  summary: This article discusses effective goal-setting and metric selection by distinguishing
    between leading and lagging metrics. Leading metrics, which are directly influenceable,
    drive the desired outcomes reflected in lagging metrics. Outcome goals, often
    tied to these lagging metrics, can lead to dissatisfaction even when achieved.
    Instead, the focus should be on process goals linked to leading metrics, fostering
    habits that naturally lead to long-term success. Examples such as pottery, UFC
    training, and business growth illustrate the importance of process-oriented over
    outcome-oriented approaches, highlighting the fulfillment and clarity that come
    from this mindset shift. Key strategies include measuring and adjusting leading
    metrics and enjoying the journey through consistent process goals.
writing/posts/hands-part-1.md:
  hash: 720db136728dc8ffa7d98289bc07b9e1
  summary: 'In "Losing My Hands," Jason, a machine learning engineer, shares his personal
    journey of struggling with severe hand pain during the COVID-19 pandemic due to
    overworking. Earning a high salary and thriving professionally, he became consumed
    by work, leading to debilitating hand cramps and the ability to perform daily
    tasks, ultimately forcing him to take a two-year hiatus. He reflects on the societal
    pressures of deriving self-worth from productivity, undergoing various treatments,
    and his ongoing recovery process. Emphasizing a resilient mindset, he now focuses
    on open-source projects and consulting, aiming to maintain a healthier work-life
    balance. Keywords: machine learning, overworking, work-life balance, injury, productivity,
    self-worth, pandemic, resilience.'
writing/posts/hands-part-2.md:
  hash: a63f46958f86748785dd7b79d1fec578
  summary: 'In "Losing My Hands Pt. 2: The Two Jasons," the author reflects on the
    impact of a repetitive strain injury (RSI) during the COVID-19 pandemic on their
    personal values and identity. The narrative explores the author''s upbringing
    as a child of immigrants, emphasizing the influence of traditional values like
    stoicism and utility while grappling with modern expectations. Through introspective
    storytelling, the author examines the dichotomy between their past and present
    selves, discussing existentialism and the importance of relationships and personal
    growth over purely financial or professional achievements. Keywords: RSI, COVID-19
    pandemic, existentialism, immigrant experience, personal growth, traditional values,
    modern identity, introspection.'
writing/posts/hands-part-3.md:
  hash: 29dfd6193b7a209a8e31f15eb632385a
  summary: 'This blog entry, titled "Losing My Hands, Advice and Tips for RSI," details
    the author''s battle with tendonitis in their wrists and thumbs, caused by overworking
    during COVID-19. It outlines significant life changes such as transitioning to
    full-time consulting, reevaluating personal values, and embracing writing as a
    form of expression. The post provides practical advice for managing RSI, including
    the use of specific braces, exercises, contrast therapy, and ergonomic devices
    like the Kinesis Pro keyboard and a vertical mouse. Emphasizing the importance
    of rest for recovery, the author shares insights into leveraging skills beyond
    physical capabilities and highlights the importance of continuous learning and
    teaching. Keywords: RSI, tendonitis, ergonomic devices, rest, Kinesis Pro keyboard,
    consulting, writing.'
writing/posts/hiring.md:
  hash: 2212ab61fca965480c8b59f2821e0eac
  summary: 'The article "Hiring MLEs at Early-Stage Companies" discusses common mistakes
    startups make when hiring machine learning engineers (MLEs) too early in their
    development process. It argues that hiring specialized MLEs when the focus should
    be on app development and integration often leads to these professionals performing
    non-core tasks. The piece emphasizes the importance of having a full-stack AI
    engineer in initial stages and highlights the lack of dedicated research teams
    as a hindrance to AI progress. It suggests that startups should focus on building
    robust infrastructure and integration capabilities before bringing in MLEs to
    maintain momentum and efficiency. Keywords: early-stage startups, machine learning
    engineers, app development, integration, AI progress, research teams, infrastructure,
    startup hiring practices.'
writing/posts/how-to-write.md:
  hash: f94a45263e29f82f365f1a82f8642d5c
  summary: 'This article outlines a strategic approach to writing compelling and effective
    content, emphasizing strong titles, evidence-based claims, reader-focused perspectives,
    and engaging introductions. It advises against using adjectives in favor of quantifiable
    data and stresses the importance of connecting with the reader''s emotions and
    needs. The piece highlights techniques such as using lists, foreshadowing content,
    and creating strong hooks to retain and engage readers. Additionally, it underscores
    the significance of positioning oneself as a knowledgeable leader or oracle and
    encourages the inclusion of a clear call-to-action (CTA). Core ideas focus on
    audience engagement, emotional connection, and providing tangible value, aiming
    to help writers craft impactful content. Keywords: content writing, titles, evidence-based
    claims, reader engagement, emotional connection, call-to-action, audience-focused
    writing.'
writing/posts/kojima-sticks.md:
  hash: b35705839e85dcdcc73b17d7510149e0
  summary: The article explores Hideo Kojima's philosophy of using 'ropes' instead
    of 'sticks' in game design to empower users, and applies this to the use of Large
    Language Models (LLMs). It argues that LLMs should go beyond simple tasks like
    answering questions or content generation to foster critical thinking and creativity,
    thus serving as catalysts for intellectual growth. By shifting from passive content
    delivery to active learning, LLMs can revolutionize user engagement. The text
    advocates for incorporating Kojima's philosophy into LLMs to enhance their potential
    in education and personal development, encouraging users to explore and engage
    more deeply with content. Key concepts include empowerment, creative thinking,
    user engagement, and the transformation of LLM applications.
writing/posts/learning.md:
  hash: 434905558f237b137b87dacce470fc2c
  summary: The blog post "Learning to Learn" by jxnl explores the author's personal
    learning journey, illustrating effective strategies for acquiring new skills.
    Key concepts include the importance of cultivating a distraction-free environment,
    prioritizing process-based goals over outcomes, and leveraging teaching to reinforce
    learning. The author emphasizes "periodizing training" to manage learning volume
    effectively and advocates learning from actively engaged experts. The post also
    highlights the significance of mastering fundamentals to avoid unforced errors,
    with examples drawn from various disciplines like design, data science, pottery,
    weightlifting, and Rocket League. These insights aim to guide readers in developing
    a structured approach to skill acquisition and personal growth.
writing/posts/life-millions.md:
  hash: cb38c5084d6337781ab0e856eb43e478
  summary: 'This article explores a thought experiment on how life might change with
    $20 million, focusing on financial freedom and its impact on lifestyle choices.
    Key aspects include optimizing work-life balance by working on passion projects,
    creating an inspiring home environment, and dedicating time to activities like
    pottery, Brazilian Jiu-Jitsu, swimming, and spearfishing. Travel plans emphasize
    intentional, enriching experiences with quarterly trips and international adventures.
    The author emphasizes quality over quantity, with a focus on personal growth,
    learning, and fostering meaningful relationships. The exercise helps clarify priorities,
    underscoring that many elements of an ideal life are attainable through intentional
    decisions, regardless of wealth. Keywords: financial freedom, lifestyle choices,
    work-life balance, personal growth, intentional living.'
writing/posts/llmops.md:
  hash: 50db587a3bf3c30bd1dda859f60f6540
  summary: "This article, \"Good LLM Observability is Just Plain Observability,\"\
    \ explores the application of traditional observability tools in enhancing large\
    \ language models (LLMs) and AI agents. By integrating Open Telemetry\u2014a suite\
    \ of APIs and tools for generating and analyzing telemetry data\u2014agents can\
    \ achieve improved performance, transparency, and self-optimization. The article\
    \ highlights the importance of comprehensive telemetry, moving beyond superficial\
    \ dashboard insights to deeply analyze LLM operations and refine agent behavior.\
    \ Key examples include using telemetry for evaluating and fine-tuning AI models,\
    \ such as a GPT-3.5 model enhanced by GPT-4 outputs. This approach bridges AI\
    \ and distributed systems, facilitating self-improvement and robust incident detection."
writing/posts/mvp.md:
  hash: d28d41bcce8bdab307055f17797e2f9b
  summary: 'This article, titled "A Feat of Strength MVP for AI Apps," discusses how
    to effectively ship a minimum viable product (MVP) for AI applications. It emphasizes
    understanding the Pareto principle in AI contexts, focusing on serving well-defined
    user segments, and the importance of educating users about the application''s
    limitations. Key points include recognizing the limitations of AI systems beyond
    summary statistics, learning to reject certain tasks, and prioritizing feature
    development based on user needs. The article highlights the benefits of targeting
    specific niches, obtaining rapid user feedback, and iterating quickly to enhance
    the product''s performance and trustworthiness. Keywords: AI applications, MVP,
    Pareto principle, user segments, feedback, niche targeting.'
writing/posts/natrualistic-diaglogue.md:
  hash: e4c6d83228233506c70753abbd10bcd1
  summary: 'The post "Implementing Naturalistic Dialogue in AI Companions" explores
    how AI can improve conversational realism by mimicking naturalistic dialogue,
    characterized by contextual efficiency, implicit references, fragmentation, and
    organic flow. Key challenges include AI''s tendency to over-explain due to its
    reliance on Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs),
    which often prioritize explicit communication. The author proposes enhancing AI
    dialogue through a memory distillation process and a response planning algorithm,
    enabling AI to use chain-of-thought reasoning and create responses that feel spontaneous
    and human-like. Keywords: naturalistic dialogue, AI companions, contextual efficiency,
    implicit communication, LLM, RAG, memory distillation.'
writing/posts/prompt-optimization.md:
  hash: 42de209c9053ced50abb15c2445df98d
  summary: The article "What is prompt optimization?" by jxnl explores the concept
    of prompt optimization in the context of language models (LLMs) and how it enhances
    content generation by refining prompts with few-shot learning examples. This process
    involves treating few-shot examples as hyperparameters, experimenting with various
    combinations to improve output quality. A significant challenge mentioned is finding
    a reliable function to score outputs, vital for evaluating effectiveness in production
    settings. The article highlights the importance of strategic example selection
    and the difficulty of quantifying subjective qualities like motivation or persuasion
    in generated content. Keywords include prompt optimization, few-shot learning,
    language models, hyperparameters, and output scoring.
writing/posts/rag-flywheel.md:
  hash: 1aa0f62ab56e59a5a35747b1ed979b24
  summary: 'This article outlines a systematic approach for building and improving
    Retrieval-Augmented Generation (RAG) systems, emphasizing the importance of prioritizing
    search accuracy before generation. It introduces a nine-step flywheel process:
    initial implementation, synthetic data generation, fast evaluations, real-world
    data collection, classification and analysis, system improvements, production
    monitoring, user feedback integration, and iteration. Key strategies include using
    synthetic data for baseline evaluations, focusing on leading metrics like precision
    and recall, and implementing fast, unit test-like evaluations. The approach also
    emphasizes real-world data clustering and a continuous improvement loop to adapt
    to changing user needs. This RAG playbook ensures data-driven product decisions
    and ongoing enhancement of AI-driven systems. Keywords: RAG systems, synthetic
    data, retrieval accuracy, continuous improvement, data-driven decisions.'
writing/posts/rag-improving-rag.md:
  hash: c5a44f56e8951553518c7e253bb2c7a3
  summary: This article offers a comprehensive guide to systematically improving Retrieval-Augmented
    Generation (RAG) systems, presenting practical strategies for developers and organizations.
    Key areas covered include the use of synthetic data for precision and recall evaluation,
    combining full-text and vector searches for optimal retrieval, implementing effective
    user feedback mechanisms, and leveraging metadata. The guide emphasizes the importance
    of clustering user queries to identify issues and continuously monitoring and
    experimenting with system performance to make informed improvements. Additionally,
    it discusses the balance between latency and performance based on specific use
    cases.
writing/posts/rag-inverted.md:
  hash: 74b54f742d943334f26b730fdb2bd713
  summary: The article discusses the concept of inverted thinking to illustrate how
    not to build a Retrieval-Augmented Generation (RAG) system, emphasizing the importance
    of considering a RAG system primarily as a recommendation platform. Key points
    include avoiding latency issues, displaying intermediate results, not showing
    source documents, ignoring user churn, and avoiding customization in UI, search,
    and embeddings. It stresses the need for manual curation of inventory, analysis
    of inbound queries, and the involvement of machine learning engineers in data
    ingestion. The article humorously critiques the neglect of custom pipelines, clean
    data requests, clustering, and the reliance on local evaluations over A/B tests,
    highlighting the pitfalls these practices can lead to.
writing/posts/rag-levels-of-rag.md:
  hash: e526b7ad2288dcd0f5f2b7bafac88143
  summary: 'This comprehensive guide explores the different levels of complexity in
    implementing Retrieval-Augmented Generation (RAG) applications. It starts with
    the basics, introducing foundational concepts such as text chunking, embedding,
    and database storage. As it progresses to more advanced levels, it covers improved
    asynchronous programming, advanced search and ranking algorithms, and structured
    response generation. The guide also emphasizes the importance of observability
    and logging for performance monitoring, evaluations using both synthetic and real
    data, and understanding application shortcomings through data analysis. Advanced
    topics like query enhancement and summarization techniques are also mentioned
    as future areas of exploration. Keywords: RAG applications, text processing, embedding,
    search optimization, observability, evaluations, query enhancement, summarization.'
writing/posts/rag-lgtmk.md:
  hash: 2f2be9b1624d8c3f5d6053e48071b6da
  summary: This blog post by Jxnl critiques the use of "LGTM@Few" as a metric for
    evaluating retrieval-augmented generation (RAG) systems, arguing it's ineffective.
    The post advises on when to rely on data over intuition, emphasizes the importance
    of velocity in testing and iterating metrics, and discusses several evaluative
    metrics like Mean Average Recall (MAR), Mean Average Precision (MAP), Mean Reciprocal
    Rank (MRR), and Normalized Discounted Cumulative Gain (NDCG). It highlights the
    need for metrics to align with business outcomes such as user satisfaction, engagement,
    conversion, and retention, and warns against pitfalls like Simpson's Paradox.
    Key points include balancing fast metrics with business goals and understanding
    the relevancy and ranking of text chunks to enhance system performance.
writing/posts/rag-low-hanging-fruit.md:
  hash: 6edfb2fa5b23057df4297f424838877b
  summary: This blog post explores strategies for enhancing RAG (Retrieval-Augmented
    Generation) search systems, providing practical tips for improving search accuracy
    and relevance. Key areas of focus include using synthetic data to establish baseline
    metrics, implementing date filters for more relevant results, refining user feedback
    mechanisms, tracking metrics like average cosine distance and Cohere reranking
    scores, and combining full-text with semantic search. Additionally, it advocates
    for restructuring text chunks into question-like formats and incorporating file
    and document metadata to optimize search results. These insights aim to improve
    RAG systems' performance, providing more accurate and timely information to users.
writing/posts/rag-plusplus.md:
  hash: a0b6c6fb2bc60bfb83c172e7cf7e2bca
  summary: In the article "Predictions for the Future of RAG," the authors discuss
    the evolving role of Retrieval-Augmented Generation (RAG) systems, predicting
    a shift from question-answering applications to report generation over the next
    6 to 8 months. This transformation is driven by the higher value reports provide
    as decision-making tools, compared to the limited scope of current RAG systems.
    The article emphasizes the importance of using Standard Operating Procedures (SOPs)
    to standardize report formats, ensuring outcomes align with business objectives.
    The authors foresee a future where report-generating tools populate marketplaces,
    maximizing RAG's potential for strategic decision-making. Keywords include RAG,
    report generation, decision-making, SOPs, and value.
writing/posts/rag.md:
  hash: 5a81db08d443426b17fec3843ef24b32
  summary: 'The article discusses the limitations of the basic Retrieval Augmented
    Generation (RAG) models that rely solely on embedding search in vector databases,
    labeling this approach as "demoware." It introduces more advanced RAG models that
    incorporate query understanding to improve precision and recall, allowing for
    nuanced and tailored searches across multiple backends. Two case studies are presented:
    Metaphor Systems, which optimizes natural language queries for search, and a personal
    assistant that dispatches queries to different backends like email and calendar.
    The article highlights the use of tools like Pydantic and the instructor library
    to model and execute complex queries, aiming to demonstrate how an efficient RAG
    system can be implemented. Key terms include RAG, large language models (LLMs),
    query understanding, vector databases, and search optimization.'
writing/posts/recsys-frameworks.md:
  hash: 7fe2447e32c2b4c04009d25fa35e1217
  summary: Stitch Fix's "Flight" framework revolutionizes the management of data pipelines,
    aiming to streamline recommendation systems by providing a modular and efficient
    architecture. Flight acts as a semantic bridge for data scientists, enabling business
    users to define pipelines in plain English, while offering robust logging, debugging,
    and tracing capabilities. Key features include error handling, distributed tracing
    with OpenTelemetry, dynamic execution, and detailed debugging tools. By implementing
    modular operator classes for data transformations, Flight enhances reliability
    and performance monitoring in real-time systems, crucial for efficient and scalable
    data pipeline operations.
writing/posts/self-loathing.md:
  hash: 7271b84624939425fa4535bc22f6cb45
  summary: "This piece reflects on personal experiences of overcoming self-loathing\
    \ and resentment towards wealth and privilege, authored by someone who once despised\
    \ rich people due to a challenging upbringing. The narrative delves into struggles\
    \ with poverty, familial relationships, and the drive to achieve financial security\
    \ while grappling with personal injuries and self-worth. Key themes include the\
    \ journey of choosing happiness, overcoming impostor syndrome, and understanding\
    \ individual suffering without comparison. Insight is drawn from the writer's\
    \ experiences, aiming to help others\u2014including the author's sister\u2014\
    with advice on resilience and self-acceptance. The article also highlights reactions\
    \ from various audiences, illustrating diverse interpretations of the writer's\
    \ advice. The message is punctuated with a quote from Naval, emphasizing that\
    \ resenting wealth can prevent its attainment."
writing/posts/self-reflections-hiring.md:
  hash: 7b190400dc4cf378b6380e23d474e273
  summary: 'This self-reflection essay explores personal growth and success, inspired
    by Graham Duncan''s insights on understanding oneself and others. Key themes include
    the importance of ownership in work, a drive for simplified and high-leverage
    decision-making frameworks, and the ability to prioritize effectively. The author
    reflects on feedback regarding work-life balance and emphasizes the value of being
    open to change and admitting mistakes. Significant growth is noted in adapting
    to challenges and finding leverage in professional projects, with a focus on people
    who exhibit obsession and curiosity as being exceptional colleagues. Keywords:
    self-reflection, personal growth, success, decision-making, leverage, work-life
    balance, ownership, curiosity.'
writing/posts/stochastic-software.md:
  hash: acdaa1f305bc5326d6d82f82c0a2d974
  summary: 'The article "Tips for Probabilistic Software" offers guidance for junior
    software engineers transitioning into machine learning, focusing on decision-making
    under uncertainty. It emphasizes understanding the nature of probabilistic systems
    and provides critical questions to guide developers: the utility of experiments,
    measurement of specific improvements, the decision-making utility of results,
    and conditions for reevaluation. Key insights include managing the long tails
    of probabilistic systems versus deterministic edge cases, designing effective
    experiments and metrics, and making focused decisions. The article highlights
    the importance of learning from both positive and negative results, documenting
    outcomes, and fostering team learning. Keywords include probabilistic software,
    machine learning, decision-making, experiments, metrics, edge cases, long tails,
    and junior software engineers.'
writing/posts/strings.md:
  hash: fb1d1c28cc57ea945b8c55e46c70f070
  summary: The article criticizes the complexity added by many LLM libraries when
    formatting prompts, advocating for a simpler approach to interact with APIs like
    OpenAI by directly using arrays of messages. The author highlights issues with
    libraries that force users to format prompts as string blocks or use convoluted
    methods like manipulating docstrings, which can limit flexibility in prompt engineering.
    The piece emphasizes the benefits of minimal abstraction, allowing immediate use
    of new API features such as additional message types, image URLs, and Base64 encoded
    images, without waiting for library updates. The author supports a straightforward,
    extensible approach where developers can easily integrate new capabilities and
    handle complex interactions like function or tool call responses.
writing/posts/techno-skepticism.md:
  hash: 9ea1907588af62d8054a3b71a224bdb4
  summary: The article "Unraveling the History of Technological Skepticism" explores
    the recurring theme of skepticism towards new technologies, using examples like
    the telephone and calculators. It highlights Marshall McLuhan's Tetrad of Media
    Effects to analyze the impact of technology, emphasizing that innovations, including
    AI, should be seen as collaborative tools rather than threats. Concepts like Centaur
    Chess illustrate AI's potential to augment human abilities. The article underscores
    AI's role in enhancing skills without replacing human insight, advocating for
    a balanced approach in integrating AI into education and society. Key terms include
    technological skepticism, AI collaboration, McLuhan's Tetrad, and Centaur Chess.
writing/posts/testosterone.md:
  hash: 5887e5332e2bff61a7ee1d48266915ea
  summary: In the blog "My year at 1100ng/dL," the author shares their personal experience
    of doubling testosterone levels from 500 ng/dl to 1100 ng/dl over a year through
    a combination of supplements, diet, and intense exercise. They discuss the motivations
    behind this experiment, influenced by the likes of Andrew Huberman and David Sinclair,
    and the impact it had on their physical and mental state, including increased
    motivation and energy but also altered perceptions of others. The author emphasizes
    the importance of understanding testosterone levels for every man, while cautioning
    that the extreme drive and mindset induced may not be sustainable long-term. Key
    elements include testosterone, supplements, motivation, health, and personal development.
writing/posts/topics_and_capabilities.md:
  hash: 42ab5d572dd7cafea5083ba741b8662f
  summary: 'The article "Art of Looking at RAG Data" emphasizes improving RAG (Retrieval-Augmented
    Generation) applications by focusing on two main aspects of search systems: topics
    and capabilities. "Topics" relate to content coverage, ensuring that user queries
    can be answered by having the necessary information in the database. "Capabilities"
    involve the effective retrieval and presentation of this information, facilitated
    through proper metadata and indexing. Examples from industries like Netflix and
    food delivery services illustrate how identifying topic gaps and enhancing search
    capabilities can drive improvements. Key objectives include content relevance,
    efficient search functionality, and minimizing data mismatches through strategic
    content and metadata management.'
writing/posts/trade-off-tool-selection.md:
  hash: 26cbe2f97abc9233ff6c3d1e055eb11d
  summary: 'The article "Optimizing Tool Retrieval in RAG Systems: A Balanced Approach"
    discusses strategies for improving tool selection in Retrieval-Augmented Generation
    (RAG) systems by balancing recall and precision. It highlights the challenge of
    selecting effective tools among a large set, emphasizing the trade-off between
    including all tools for high recall versus optimizing precision to enhance system
    performance. The article introduces the "Evergreen Tools" strategy, which involves
    identifying and consistently using the top 20% of tools that handle most use cases
    while employing retrieval methods for the less frequently used tools. By analyzing
    usage data, creating a histogram of tool usage, and understanding the Pareto principle,
    system efficiency and precision are improved. The piece underscores the need for
    ongoing analysis to adapt to changing user needs. Key terms include tool retrieval,
    RAG systems, precision, recall, and data-driven optimization.'
writing/posts/wandb-course.md:
  hash: bc523d3c49018174f804ad7f4dfe6b04
  summary: Explore the free "Weights and Biases" course, designed for anyone interested
    in learning about steering language models. Created by JXNL and available at [wandb.courses](https://www.wandb.courses/courses/steering-language-models),
    this course is accessible, concise, and under an hour long. Dive into the fundamentals
    of managing and understanding language model behaviors, with practical insights
    and techniques. Perfect for those looking to enhance their skills in machine learning
    and artificial intelligence using Weights and Biases tools.
writing/posts/whoami.md:
  hash: 4647ba6726493100d741e0620496354f
  summary: This blog post introduces the author, who plans to share a blend of technical
    machine learning insights and personal reflections in the upcoming year. The author,
    originally from a rural Chinese village, delved into art, science, and eventually
    machine learning, with pivotal experiences at Waterloo University, Facebook, and
    Stitch Fix. A significant personal challenge arose with a diagnosis of RSI in
    2020, prompting a reevaluation of self-worth away from productivity. Embracing
    new interests like martial arts, pottery, and writing, the author now advises
    startups and seeks to capture thoughts in writing as a form of personal fulfillment.
    Key themes include machine learning, personal growth, career transitions, overcoming
    adversity, and self-discovery.
writing/posts/writing-distribution.md:
  hash: 968d3a73995ad7baf7ebf68cc0f69e23
  summary: This article discusses the challenges and unexpected consequences of online
    recognition as a data scientist. Initially motivated by the desire to build a
    professional network and visibility for speaking engagements, the author shares
    their experience of being recognized in public, which has transitioned from exciting
    to overwhelming. Highlights include the impact of being frequently approached
    during industry events, disruption of personal and professional time, and the
    need for strategic public appearances. The article underscores the double-edged
    sword of online influence, emphasizing the importance of distribution and adapting
    to the real-world implications of digital success. Key topics include online presence,
    public recognition, personal challenges, and strategies for managing visibility.
writing/posts/writing-final.md:
  hash: 59e768cb613697692c4e0651bf84377a
  summary: 'This article, "Content Creation Mastery: From Tweets to Long-Form," offers
    practical strategies for crafting impactful content across various formats, from
    concise tweets to comprehensive long-form articles. Core ideas include creating
    attention-grabbing hooks and titles, using data-driven evidence over vague adjectives,
    structuring content for easy scanning, and focusing on reader-centric writing.
    It emphasizes the importance of including clear calls-to-action and iterating
    content based on data analytics. Keywords include content creation, writing, marketing,
    tweets, long-form content, SEO, and audience engagement.'
writing/posts/writing-style-guide.md:
  hash: e36f0c24b271a4c9a59952bf5a072e43
  summary: '"Content Creation Mastery: 9 Strategies to 10x Your Impact" provides actionable
    insights for creating impactful content. Key strategies include crafting attention-grabbing
    titles, using powerful intros, providing evidence over adjectives, and foreshadowing
    value. The guide emphasizes structuring content for easy skimming, focusing on
    audience needs, predicting future trends, offering a clear call-to-action, and
    iterating based on data. Core ideas revolve around enhancing engagement by addressing
    reader pain points, being specific, and ensuring content is both compelling and
    data-driven. Keywords: content creation, audience engagement, data-driven strategies,
    impactful content, SEO, call-to-action.'
writing/posts/writing-tweet.md:
  hash: c8f14f9f3a8b674081f432dbca725898
  summary: The article "Anatomy of a Tweet" explores strategies for crafting effective
    tweets to grow a Twitter following, focusing on writing compelling hooks, engaging
    content, and strong calls to action. Key points include the importance of creating
    a clear benefit for readers, using lists or metrics, and maintaining relevance
    and engagement throughout the tweet. Additionally, the article emphasizes the
    significance of a call to action to maximize conversions, such as requesting follows
    or shares. The author also highlights using Twitter for generating business leads
    and testing content ideas. Keywords include Twitter, tweet writing, hooks, engagement,
    calls to action, social media strategy, and follower growth.