R&D Notes

Engineering Logs

Building the infrastructure for sovereign AI. Technical notes on what works, what breaks, and how we fix it.

DATA ENGINEERING

Building an Automated RAG Data Pipeline: PDF to Fine-Tuning Ready

The architecture behind doc2dataset

A deep dive into building a Python pipeline that transforms messy PDFs and unorganized docs into high-quality datasets for Llama-3 fine-tuning and RAG.

PythonRAG PipelineLlamaIndex+2

Read Log

LLM OPS

Dynamic Context Injection: Solving RAG Hallucinations

Why static vector search isn't enough for Enterprise AI

How to implement Dynamic Context Management in Python to fix the 'Lost in the Middle' phenomenon and improve RAG accuracy by 3x.

PythonHybrid SearchVector DB+2

Read Log

AGENT INFRASTRUCTURE

Automating Agent Infrastructure: OpenAPI to Function Calling

Zero-config tool generation for AI Agents

Stop writing manual tool definitions. How we built a TypeScript SDK that parses OpenAPI specs and auto-generates function-calling definitions for OpenAI and Claude.

TypeScriptOpenAPIFunction Calling+2

Read Log