writings
- may 2026 three flavors of simulating an agent pipeline
- may 2026 evolving an eval harness from one bash script to a merge gate
- oct 2025 chunk twice, retrieve once: double-pass semantic chunking
- sep 2025 building a context system that agents can actually use
- jul 2025 teaching a small llm to filter its own context
- jun 2025 grpo + lora: making a 3b model reason
- jun 2025 colqwenrag: rag without ocr, for legal papers
- jun 2025 fine-tuning embeddings until legal-rag stops missing
- mar 2025 rolling your own reasoner: grpo + sft from scratch
- mar 2025 how to make a model think: the deepseek recipe
- mar 2025 a llm that drafts indian legal contracts