Shipping AI that works
Forward Deployed Engineer. I work where the customer's problem meets the platform.
Where I am now
I'm a Forward Deployed Engineer at Echos, building AI and data solutions on Palantir Foundry and AIP. My job is to turn complex, messy enterprise data into reliable workflows that support real business decisions, and to stay with the problem until it works in production.
Before that I co-founded Companion Labs and took Voxerra, an AI voice receptionist for Indian clinics, from customer conversations to a live Google Play app: 11 Indian languages, sub-800ms latency, and a speech pipeline 93% cheaper than where we started.
I spent my first five years at Honeywell, shipping field-inspection software used by thousands of workers across enterprise deployments. I studied at IIT Kharagpur and I'm a Claude Certified Architect.
Problem First
Start from the customer's actual problem, not the platform. Vague requirements become a clear architecture before any code ships.
Trustworthy AI
Pair LLMs with deterministic, rule-based checks so results stay accurate and explainable once they reach production.
Ship to Production
Stay with a problem until it works for the people using it, not just in a demo.
