»» auto mode on · esc to interrupt · 14 posts in context
Alex Jacobs — Blog
Alex Jacobs
AI Systems Integration | Data Science | Full-Stack Development | DevOps | Bioinformatics
1# Alex Jacobs
2
3AI Systems Integration | Data Science | Full-Stack Development | DevOps | Bioinformatics
4
5<!-- ~/alex-jacobs.com · 14 posts in workspace · utf-8 · lf -->
6
7## Recent files
8- [I Am Morally Opposed to Updating My CLAUDE.md](~/posts/claudemd/)
9- [42 Is Not a Random Number](~/posts/42-is-not-a-random-number/)
10- [Beating BERT? Small LLMs vs Fine-Tuned Encoders for Classification](~/posts/beatingbert/)
11
12**[open the full workspace](~/posts/)**
Latest posts
I Am Morally Opposed to Updating My CLAUDE.md
Every CLAUDE.md accumulates rules written in frustration after a single bad agent run, and never loses them. Why agent instruction files decay into a grievance archive, and what to do instead.
Read(I Am Morally Opposed to Updating My CLAUDE.md)
Every CLAUDE.md accumulates rules written in frustration after a single bad agent run, and never loses them. Why agent instruction files decay into a grievance archive, and what to do instead.
I Am Morally Opposed to Updating My CLAUDE.md
Every CLAUDE.md accumulates rules written in frustration after a single bad agent run, and never loses them. Why agent instruction files decay into a grievance archive, and what to do instead.
I Am Morally Opposed to Updating My CLAUDE.md.md
2026-08-19Every CLAUDE.md accumulates rules written in frustration after a single bad agent run, and never loses them. Why agent instruction files decay into a grievance archive, and what to do instead.
42 Is Not a Random Number
MCP won and agents are everywhere, but they still fail in the same dumb ways. The fix isn't a smarter model. It's deterministic tools.
Read(42 Is Not a Random Number)
MCP won and agents are everywhere, but they still fail in the same dumb ways. The fix isn't a smarter model. It's deterministic tools.
42 Is Not a Random Number
MCP won and agents are everywhere, but they still fail in the same dumb ways. The fix isn't a smarter model. It's deterministic tools.
42 Is Not a Random Number.md
2026-02-08MCP won and agents are everywhere, but they still fail in the same dumb ways. The fix isn't a smarter model. It's deterministic tools.
Beating BERT? Small LLMs vs Fine-Tuned Encoders for Classification
I ran 32 experiments comparing small LLMs to BERT on classification tasks. Turns out 2018-era BERT is still really good at what it does.
Read(Beating BERT? Small LLMs vs Fine-Tuned Encoders for Classification)
I ran 32 experiments comparing small LLMs to BERT on classification tasks. Turns out 2018-era BERT is still really good at what it does.
Beating BERT? Small LLMs vs Fine-Tuned Encoders for Classification
I ran 32 experiments comparing small LLMs to BERT on classification tasks. Turns out 2018-era BERT is still really good at what it does.
Beating BERT? Small LLMs vs Fine-Tuned Encoders for Classification.md
2026-01-04I ran 32 experiments comparing small LLMs to BERT on classification tasks. Turns out 2018-era BERT is still really good at what it does.
The Case Against pgvector
The real limitations of pgvector in production: index choice between IVFFlat and HNSW, why real-time search is hard, pre- vs post-filtering pain, how performance degrades at scale, and when a dedicated vector database is the better call.
Read(The Case Against pgvector)
The real limitations of pgvector in production: index choice between IVFFlat and HNSW, why real-time search is hard, pre- vs post-filtering pain, how performance degrades at scale, and when a dedicated vector database is the better call.
The Case Against pgvector
The real limitations of pgvector in production: index choice between IVFFlat and HNSW, why real-time search is hard, pre- vs post-filtering pain, how performance degrades at scale, and when a dedicated vector database is the better call.
The Case Against pgvector.md
2025-10-29The real limitations of pgvector in production: index choice between IVFFlat and HNSW, why real-time search is hard, pre- vs post-filtering pain, how performance degrades at scale, and when a dedicated vector database is the better call.
A Production Framework for LLM Feature Evaluation
Which LLM application patterns actually survive production, based on extraction, generation and classification features shipped to real users. Includes what to evaluate before building and what reliably fails.
Read(A Production Framework for LLM Feature Evaluation)
Which LLM application patterns actually survive production, based on extraction, generation and classification features shipped to real users. Includes what to evaluate before building and what reliably fails.
A Production Framework for LLM Feature Evaluation
Which LLM application patterns actually survive production, based on extraction, generation and classification features shipped to real users. Includes what to evaluate before building and what reliably fails.
A Production Framework for LLM Feature Evaluation.md
2025-06-01Which LLM application patterns actually survive production, based on extraction, generation and classification features shipped to real users. Includes what to evaluate before building and what reliably fails.
A Computer Made This
OpenAI's 4o generates images by reasoning in pixel space rather than diffusing noise, which is why it can render legible text and follow multi-step edits. What that shifts for creative work.
Read(A Computer Made This)
OpenAI's 4o generates images by reasoning in pixel space rather than diffusing noise, which is why it can render legible text and follow multi-step edits. What that shifts for creative work.
A Computer Made This
OpenAI's 4o generates images by reasoning in pixel space rather than diffusing noise, which is why it can render legible text and follow multi-step edits. What that shifts for creative work.
A Computer Made This.md
2025-03-27OpenAI's 4o generates images by reasoning in pixel space rather than diffusing noise, which is why it can render legible text and follow multi-step edits. What that shifts for creative work.
RAG: From Context Injection to Knowledge Integration
A technical dive into the limitations of current RAG approaches, examining architectural challenges and exploring pathways to more integrated knowledge-aware LLM architectures.
Read(RAG: From Context Injection to Knowledge Integration)
A technical dive into the limitations of current RAG approaches, examining architectural challenges and exploring pathways to more integrated knowledge-aware LLM architectures.
RAG: From Context Injection to Knowledge Integration
A technical dive into the limitations of current RAG approaches, examining architectural challenges and exploring pathways to more integrated knowledge-aware LLM architectures.
RAG: From Context Injection to Knowledge Integration.md
2025-02-17A technical dive into the limitations of current RAG approaches, examining architectural challenges and exploring pathways to more integrated knowledge-aware LLM architectures.
Python Async Programming: A Deep Dive into async/await and the Event Loop
How async Python actually works: what async def and await do, how the asyncio event loop schedules coroutines, tasks and futures explained, and when to use async instead of threads or processes.
Read(Python Async Programming: A Deep Dive into async/await and the Event Loop)
How async Python actually works: what async def and await do, how the asyncio event loop schedules coroutines, tasks and futures explained, and when to use async instead of threads or processes.
Python Async Programming: A Deep Dive into async/await and the Event Loop
How async Python actually works: what async def and await do, how the asyncio event loop schedules coroutines, tasks and futures explained, and when to use async instead of threads or processes.
Python Async Programming: A Deep Dive into async/await and the Event Loop.md
2024-01-28How async Python actually works: what async def and await do, how the asyncio event loop schedules coroutines, tasks and futures explained, and when to use async instead of threads or processes.
Integration Testing FastAPI: Mocking Auth, MongoDB, S3, and External APIs
A practical guide to FastAPI integration tests with pytest: mocking JWT authentication and the TestClient, patching external API calls, and faking MongoDB and S3 with mongomock and moto.
Read(Integration Testing FastAPI: Mocking Auth, MongoDB, S3, and External APIs)
A practical guide to FastAPI integration tests with pytest: mocking JWT authentication and the TestClient, patching external API calls, and faking MongoDB and S3 with mongomock and moto.
Integration Testing FastAPI: Mocking Auth, MongoDB, S3, and External APIs
A practical guide to FastAPI integration tests with pytest: mocking JWT authentication and the TestClient, patching external API calls, and faking MongoDB and S3 with mongomock and moto.
Integration Testing FastAPI: Mocking Auth, MongoDB, S3, and External APIs.md
2023-06-11A practical guide to FastAPI integration tests with pytest: mocking JWT authentication and the TestClient, patching external API calls, and faking MongoDB and S3 with mongomock and moto.
CheeseGPT
A complete walkthrough of a minimal RAG pipeline in Python: scraping a corpus, chunking it, creating embeddings, storing them in Redis, and answering questions over the result with GPT.
Read(CheeseGPT)
A complete walkthrough of a minimal RAG pipeline in Python: scraping a corpus, chunking it, creating embeddings, storing them in Redis, and answering questions over the result with GPT.
CheeseGPT
A complete walkthrough of a minimal RAG pipeline in Python: scraping a corpus, chunking it, creating embeddings, storing them in Redis, and answering questions over the result with GPT.
CheeseGPT.md
2023-03-14A complete walkthrough of a minimal RAG pipeline in Python: scraping a corpus, chunking it, creating embeddings, storing them in Redis, and answering questions over the result with GPT.