skip to content

blog

thoughts and ideas

Posts

Building contextual ad infrastructure that detects conversation intent and surfaces relevant offers natively within chat interfaces — without harming user experience.

What I learned building a full-stack AI advertising platform: architecture decisions, API design, and integrating LLM-powered recommendation pipelines.

Exploring how the Model Context Protocol enables standardized tool interaction for LLMs and what it means for agentic workflow architecture.

STT + LLM + TTS pipelines, tool calling and turn detection at 92% intent accuracy.

Precision, Recall, F1, MRR and NDCG across 5+ retrieval strategies.