Builder · AI Engineering Leader
I Build AI Systems
That Work in Production
Two decades shipping production software — from FPGA design tools at QuickLogic to multi-agent AI systems. 5 products built and deployed in the last 3 years, leading architecture and full-stack delivery.
Multi-Agent Pipelines
5 Products Shipped
Two Decades of Shipping
What I build
Production-grade AI systems across three core domains.
Multi-Agent AI Systems
Orchestrated pipelines where specialized agents collaborate — content generation, data analysis, decision automation.
NL-to-SQL & Data Intelligence
Ask your database questions in plain English. Production-grade query generation with guardrails and validation.
WhatsApp & Conversational AI
Voice-to-insight pipelines, chatbots, and conversational interfaces that work on the channels your users already use.
Projects that shipped
Real products, real users, real production traffic.
Kaiwa
AI-native social dining platform — built from zero to production in under 3 months.
View Project →
AI Marketing Assistant
Autonomous branded poster generation — agents for meaning, code for pixels.
View Project →
MR Voice
Voice-to-insight for pharma field intelligence — 30-second structured reports from a voice note.
View Project →
QueryNLQ
Ask your database questions in plain English.
View Project →
How I work
No slides, no 6-month roadmaps. Build fast, ship weekly, iterate from evidence.
Discovery
I understand your problem, data, and constraints. No slides — I prototype.
Architecture
Design the system end-to-end: agents, pipelines, APIs, deployment. Ship the hard parts first.
Build & Iterate
Weekly deploys. Real users from week two. Tight feedback loops, not waterfall.
Ship & Support
Production-hardened delivery with monitoring, documentation, and long-term ownership.
Results, not promises
20+
Years shipping production software
5
Products built & deployed
<3mo
Average concept-to-production
1
Patent filed
"Too many AI projects die in the demo phase. I build the systems that make it through — the ones that handle real traffic, edge cases, and the questions nobody thought to ask during the pilot."
Engineering insights
Agents for Meaning, Code for Pixels: Building a 9-Step Poster Pipeline
How three LLM agents and six deterministic engines generate production-ready marketing posters — and why the renderer has zero AI.
Read →
Production WhatsApp AI: What the Tutorials Don't Tell You
Lessons from two WhatsApp AI systems — a stateful booking chatbot and a stateless voice-to-insight pipeline. The hard parts aren't the AI.
Read →
Ready to build something real?
I'm interested in hard technical problems — whether that's leading an AI engineering team or building a production system from scratch. Let's talk.