Intro

I've curated this list to share with those asking for my personal recommendations on what to follow regarding LLM, RAG, genAI etc.

Jason Liu (@jxnlco (opens in a new tab))

He is the maintainer of the instructor library (opens in a new tab). I use this every day because we definitely need structured outputs coming out from LLMs.

He has successfully built a consulting business for LLM/RAG apps, and he is sharing all of his process for free.

Some blog posts that has changed my process:

• When I only have a few days left to deliver something, I take inspiration from this: https://jxnl.co/writing/2024/05/11/low-hanging-fruit-for-rag-search/ (opens in a new tab)
• Getting inspirations for improving a RAG system: https://jxnl.co/writing/2024/02/28/levels-of-complexity-rag-applications/#evaluating-the-answering-system-feedback (opens in a new tab)
• New paradigm of software engineering, how to be efficient etc: https://jxnl.co/writing/2024/01/19/tips-probabilistic-software/ (opens in a new tab)

Also, he is writing content super quickly, with speech to text, + claude.ai. He is talking about that somewhere in this live Q&A (opens in a new tab).

Answer.ai (+ Jeremy Howard @jeremyphoward (opens in a new tab))

I read blog posts from answer.ai from time to time. Jeremy Howard, part of answer.ai, is also someone to follow on X.

• https://www.answer.ai/posts/2024-04-12-tips.html (opens in a new tab)
• https://www.answer.ai/posts/2024-01-26-freaktakes-lessons.html (opens in a new tab)

Hamel Husain @HamelHusain (opens in a new tab)

He has built a Twitter audience just by saying Look at your data to people who are using LLMs.

• Getting inspirations for writing evals and tests: https://hamel.dev/blog/posts/evals/#step-1-write-scoped-tests (opens in a new tab)
• LLM Free Courses: https://parlance-labs.com/education/ (opens in a new tab)
• Am I following good practices, not doing the same mistakes others did: https://applied-llms.org/ (opens in a new tab)

Eugene Yan (@eugeneyan (opens in a new tab))

Mentioned as a thoughtful writer who approaches topics from a scientific perspective. https://eugeneyan.com/writing/llm-patterns/ (opens in a new tab)

I need to read more about Eugene. I have read two of his articles by luck or randomly, so I'll just follow him in more detail.

• I like the style and choices for his excalidraw diagrams
• I like exhaustive articles

Lilian Weng (@lilianweng (opens in a new tab))

Not familiar with her writing, it's on the top of my list, so sharing it too.

Need to read:

• https://lilianweng.github.io/posts/2024-02-05-human-data-quality/ (opens in a new tab)
• https://lilianweng.github.io/posts/2024-07-07-hallucination/ (opens in a new tab)

Pieter Levels (@levelsio (opens in a new tab))

Not an AI/LLM expert, but he has built photoai.com (opens in a new tab) which is doing one of the best consumer text->img apps, replacing physical photography. He is shipping fast, out in the open. Always good to know what models / approaches he is using.

For example, he has built a NSFW filter for images using gpt-4-vision, or another one to detect bad hands. He always has good tricks because it's his only way to ship.