Learn to build real systems.
Check every claim with numbers.
Free step-by-step guides on AI engineering, the OpenAI SDK and FastAPI, plus benchmarks and experiments that ship the code to reproduce them. Built for engineers who want to know what actually works.
- 3guide series
- 66free guide parts
- 7benchmarks
- 25experiments
Guides
Step-by-step series with runnable code, diagrams and a real project at the end. Free to read, no sign-in.
Browse 3 guides →Benchmarks
Measured results for RAG, chunking, vector search and evals, with the setup and script for each.
See 7 benchmarks →Community
Tech articles and tool write-ups by engineers, reviewed by editors before they go live. Write your own.
Read the community →Guides
All guides →- LLMs · RAG · agents · MCP · observability
AI Engineering
LLMs, RAG, agents, MCP, observability and guardrails: how to build AI features that hold up in production, with runnable Python.
- The OpenAI Python SDK, end to end
OpenAI SDK
The OpenAI Python SDK end to end: Responses vs Chat, streaming, structured outputs, tools, embeddings, retries, cost and evals.
- Production web APIs
FastAPI
Production FastAPI: parameters, Pydantic, dependencies, async, databases, auth, testing, performance and deployment.
Latest benchmark results
All benchmarks →- Chunking
Chunk size sweep: 200 to 2,000 characters
How big should RAG chunks be?
0.877best strict hit@1 (heading-aware, 200 chars) - Evals
Golden-set evals in CI: catching retrieval regressions
Can a small golden set in CI catch retrieval regressions?
3 of 4injected regressions caught - Models
Predicting embeddings vs tokens with less data
Does predicting embeddings degrade more gracefully than predicting tokens?
−99.4%token-space score with 10x less data
Latest experiments
All posts →From the community
Community →Get new results by email
One short email per new experiment, plus the free 25-point RAG production checklist. Unsubscribe any time.