Blog

Writing about AI, engineering lessons, and what I'm building.

June 22, 2026#RAG#DistributedSystems#AgenticAI#FineTuning#CrossEncoder#BiEncoder#InformationRetrieval#VectorDB#VectorSearch#Reranking#SemanticSearch#LLMOps

RAG Efficiency, Part 1: Boost Retrieval Accuracy 25% With Rerankers

Cross-encoder rerankers can turn a high-recall retrieval pipeline into a precise production RAG system if you put them in the right place.

May 15, 2026AgentsLLMsEngineering

Why LLM Agents Fail in Production (And What to Do About It)

Agents look impressive in demos but break in unpredictable ways in production. Here's what I've learned about the failure modes and how to design around them.

May 7, 2026LLMsCost OptimizationEngineering

Token Cost Optimization Series, Part 1: Stop Sending JSON to Your LLM

JSON repeats field names with every record, burning tokens at scale. Here's a simple format swap that cuts context token usage by 40–58% for tabular data.

April 28, 2026RAGLLMsEngineering

RAG Beyond the Basics: What Nobody Tells You

Everyone can build a basic RAG pipeline. Making it production-ready is a different skill entirely. Here are the lessons I've learned the hard way.

March 10, 2026EvalsLLMsResearch

How to Actually Evaluate LLM Outputs

Vibes-based evaluation doesn't scale. Here's a practical framework for measuring LLM quality in ways that are reproducible, actionable, and honest.