Writing & Field Notes

Technical essays on production LLM systems, RAG architectures, observability, and agentic AI - plus in-depth field notes from building at scale.

original field notes

Open source architectures

Production RAG, evals & agentic patterns on GitHub

Technical essays

Production RAG, observability & agentic systems on Medium

Field Notes

Featured Essays on Medium

Deep technical writing on production LLM systems, RAG architectures, observability, and agentic AI.

Full archive

Feb 2026

LLM Observability: The Engineering Imperative for Production AI Systems

Traditional monitoring breaks for stochastic LLM and agentic systems. The four pillars - tracing for multi-hop RAG/agent flows, LLM-as-Judge + RAGAS evals, drift detection, and token economics - with tooling patterns that power production gates like the LLM Evaluation Harness.

ObservabilityEvalsAgenticProduction

Mar 2026

LangSmith vs Arize vs Braintrust: The Definitive 2026 Comparison

Practical decision framework for RAG and multi-agent production workloads. When to pick LangSmith for deep LangGraph tracing, Arize for enterprise drift & compliance, or Braintrust for eval-first CI - the exact trade-offs behind the open-sourced eval reference repos.

ToolingEvalsRAGCI/CD

Mar 2026

A Technical Deep Dive into Databricks AI Ecosystem

Mosaic AI, Vector Search for RAG, agent frameworks, evaluation tooling, and model serving at enterprise scale. How the patterns from ClinIQ, EvidenceIQ, and the other open-source reference repos are operationalized in real production environments.

DatabricksRAGPlatformEnterprise

Feb 2026

How to Build AI-Powered Applications

From prototype to production: growing evals from real traces, RAG/agent patterns, and shipping quality gates alongside the codebase. The methodology behind every open-source reference implementation I release on GitHub.

ProductionRAGAgentsEvals

Mar 2026

The LangChain Ecosystem

LangGraph orchestration, LangSmith tracing, and the full stack for traceable, evaluable RAG and agentic systems that ship. Directly complements the architecture decisions in the open-source reference implementations on GitHub.

LangChainLangGraphAgentsObservability

Writing & Field Notes

Field Notes

RAG failure patterns I see repeatedly

Why most GenAI demos fail in production

Evaluating LLM systems beyond accuracy

Featured Essays on Medium

LLM Observability: The Engineering Imperative for Production AI Systems

LangSmith vs Arize vs Braintrust: The Definitive 2026 Comparison

A Technical Deep Dive into Databricks AI Ecosystem

How to Build AI-Powered Applications

The LangChain Ecosystem