Plan
The request is decomposed into PRDs, BRDs, and architecture docs before a line of code exists.
Senior AI platform engineer in Hyderabad. Four years turning multi-agent systems into products that 10,000+ people actually use.

people using what I shipped
products indexed for semantic search
organizations running the work
junior engineers mentored
I joined Adya.ai as its founding engineer and started as the only person on the platform. Today the same codebase serves three organizations, and I mentor the four engineers who build on it with me.
Most of my work sits where AI systems meet real product surfaces: orchestrating agents that plan and verify each other, wiring semantic search over millions of records, then owning the frontend that has to make all of it feel simple.
Golang and Node on the back. React and React Native on the front. Agents in between.
Four years, three organizations, one long-running platform.
AI Software Engineering Platform
Natural language goes in. React, Next.js, and React Native applications come out, along with the PRDs, schemas, APIs, auth, tests, and deploy configs behind them.
Specialized agents hand work between planning, implementation, verification, review, and deployment.
OpenAI, Claude, Gemini, DeepSeek, Ollama, and OpenRouter behind a single selection layer.
Multi-tenant isolation with infrastructure provisioned automatically through Docker and Fly.io.
Dependency conflicts, hallucinated imports, failed build verification, and retry orchestration.
The request is decomposed into PRDs, BRDs, and architecture docs before a line of code exists.
Generation pipelines produce database schemas, backend APIs, auth, CRUD, and UI components.
Builds are run and checked. Hallucinated imports and dependency conflicts get caught here, not by the user.
A separate agent reads the output against the plan it was given and sends failures back around.
Docker images and Fly.io infrastructure are provisioned per workspace, isolated from every other tenant.
Frontend, backend, AI services, vector search
Semantic search across 8.8M+ products using Azure OpenAI embeddings and Weaviate. It reads intent, then recommends whole outfits, recipe-based grocery runs, and side-by-side comparisons.
Frontend architecture and performance
Buyer, seller, logistics, grievance, and RSP modules for an early ONDC technology service provider, built on an Nx monorepo with micro-frontends.
Frontend engineering
The seller-facing UI for an omnichannel commerce app inside the ONDC ecosystem. My first production surface, and the reason the rest of this list exists.
Aug 2023 to now (current)
Adya.ai
Architecting AI developer platforms and commerce systems from concept to production. Led AI adoption across the engineering organization, defined coding standards, and mentored the team inside a 20-person engineering group.
May 2023
Eunimart
Continued with the same engineering team through the transition, then converted to a full-time role that August.
Aug 2022 to Aug 2023
Eunimart
Built the Seller App UI and frontend for an omnichannel commerce application in the ONDC ecosystem using React and Redux.
2019 to 2023
Swarnandhra College of Engineering and Technology
Graduated with a CGPA of 8.0 out of 10.0.
5-Star Python Programmer
HackerRank
Certified Python Developer
CISCO
Management Appreciation Award
Swarnandhra College, for development and leadership
I read every message. Tell me what you are working on and I will tell you honestly whether I am the right person for it.