Turning ideas
into digital
reality

What I do

1. Oct 2025 – Present · Bengaluru

   Zensiblecurrent home ♥ ↙

   AI Engineer

   I co-own the core AI service: the multi-agent brain behind Zensible's Slack and Teams conversations.

   of ~1,170 commits

   792

   SQL per agent call

   58→29

   off every workflow step

   −207ms

   the test suite

   2×

   what I'm building there ↓ok, less ↑

   • ✳Runs on FastAPI, PostgreSQL and LangGraph, streaming multi-agent conversations with every customer's data isolated in the database. Halved database work per agent call: 58→29 SQL statements, 122→45 ms.
   • ✳Architected two-tier supervisor-swarm orchestration: agents hand off to peers, run in parallel with asyncio, and can't loop forever.
   • ✳Built directives end to end: admins drop a request into someone's live chat as a parallel conversation with its own memory, without derailing what they were talking about.
   • ✳Moved prompts, tools, personas, guardrails and agent structure into the database, so behaviour changes need no engineer or deploy. Config now updates instantly instead of taking up to an hour, and two paid services got retired.
   • ✳People start company workflows like leave and provisioning straight from chat; a one-shot tool makes sure each is filed exactly once.
   • ✳Built for scale: Redis atomic locks that fail open, background queues for long agent runs, size-bounded caches, and zero-retry execution so no model call is ever paid for twice.
   • ✳Wrote the team's engineering and ticket-sizing guidelines and its layered service architecture, and led a 225+ file static-analysis cleanup.

2. Sep 2023 – Oct 2025 · Kerala

   QBurst2 good years ↙

   Software Developer

   I prototyped and shipped AI for internal and client projects: agentic proofs of concept, FastAPI services, durable workflows and self-hosted LLMs.

   manual effort, via Temporal workflows

   −30%

   open-source LLMs in production on GCP

   vLLM

   of AI proofs of concept and client work

   2 yrs

   what I built there ↓ok, less ↑

   • ✳Shipped AI/LLM proofs of concept and FastAPI services, prototyping agentic solutions for internal and client projects.
   • ✳Built durable multi-stage workflow orchestration on Temporal, cutting manual effort by an estimated 30%.
   • ✳Hosted production open-source LLMs on vLLM with scalable GCP infrastructure for low-latency inference.
   • ✳Consulted on AI/LLM initiatives: architecture, model selection and agentic-flow design.

3. Jan 2023 – Sep 2023 · Ernakulam (Remote)

   Zobytwhere it started ↙

   Software Developer

   My first role: full-stack features for a social platform and a decentralized app.

   users on the APIs I built

   1,000+

   storage and retrieval latency

   −20%

   FastAPI features shipped

   10+

   what I built there ↓ok, less ↑

   • ✳Built full-stack features for Stych.social and Hop23.xyz, architecting backend services and REST APIs serving 1,000+ users.
   • ✳Developed 10+ FastAPI features and blockchain-based authentication, cutting storage and retrieval latency by 20%.

Let's build something amazing together.
Get in touch!

Build AI/full-stack solutions together.