01 // PRODUCTS & CASE STUDIES
Shipped Software & Live Systems
RAG Studio & Ingestion Engine
Upload a PDF, paste a YouTube link, or give it a website URL — the system reads it all in the background and lets you chat with that content. Every answer it gives links back to the original source, so you know it's not making things up.
Interruption-Resilient Voice Agent
A conversational voice agent that responds in under a second and handles interruptions the way a real person would — the moment you speak over it, it stops, listens, and continues without losing track of where the conversation was.
Dialogue Spark Viz
A voice agent paired with a live visual display — as you talk and as the AI responds, you can watch the audio waveform react in real time. It shows you exactly when the AI is listening, thinking, or speaking.
02 // WHAT I CAN BUILD
Product & Technical Capabilities
Autonomous AI Agents & Co-Pilots
Replacing multi-step manual processes with smart conversational tools.
I build AI agents that can actually get things done — they think through a problem, call external tools or APIs when needed, handle errors gracefully, and return structured, predictable results.
Real-Time Conversational Voice AI
Voice AI that feels natural — not robotic.
I build voice agents that sound and behave like a real person — they respond fast, pause when you interrupt them, and never lose track of the conversation.
Production RAG & Context Engineering
AI that answers from your actual data — not guesswork.
I build retrieval systems that give AI models the right information at the right time, so they answer accurately from your business data instead of making things up.
0→1 Product Architecture & Scalable Backends
Building from a blank page to millions of daily users.
From the very first database table to Dockerized deployments that handle millions of requests, I take full technical ownership — including infrastructure, scalability, and reliability.
03 // ENGINEERING PHILOSOPHY
How I Ship Fast Without Breaking Things
Write only what the problem needs. No unnecessary complexity, no bloated dependencies — just clean, purposeful code that's easy to read and even easier to change.
AI should never be a black box in a real product. Every AI output goes through strict schema validation (Zod) before it touches any application logic — no surprises in production.
I use Claude Code & Cursor rules to ship faster without cutting corners — delegating repetitive tasks to AI so I can focus on the decisions that actually require judgment.
04 // RECOGNITION
Milestones & Awards
2nd Place — Google Ener'gentic Hackathon 2025
Out of hundreds of teams, our agentic solution was selected for the Top 2 and showcased at Stanford University. We built a real-time grid energy intelligence system powered by autonomous AI agents.
Open Source Mentor at C4GT (Code for GovTech)
Selected to guide contributors building digital public goods for the Indian government — working on CORD Network (digital identity) and Sunbird-RC (open credentialing) open-source projects.
GenAI with JavaScript & Curriculum Evaluator
Completed the foundational GenAI with JavaScript cohort and was selected to review and evaluate the 2026 production curriculum rewrite — providing feedback used to improve the official course content.