Writing
In-depth articles on LLMs, RAG, KAG, and production AI engineering — written from real deployments, not theory.
Full engineering breakdown — four-stage Pub/Sub pipeline, parallel Cloud Run workers, distributed rate limiting, idempotency guarantees, and BigQuery observability. How we processed 117K leads autonomously.
The LangGraph architecture behind 117K emails, 10K AI voice calls, and $2M+ in annual revenue — what worked, what surprised me, and what I would do differently.
A practical comparison from an AI engineer who built both in production for 20,000+ legal professionals. RAG failed. KAG delivered a 40% accuracy gain.
Clinical document RAG, voice AI for scheduling, HIPAA-aware pipelines. An honest breakdown of what delivers ROI in healthcare AI and what tends to fail.