About

The person behind the systems.

AI Engineer, based in Surat, India. I like problems that only make sense once you've owned every layer of them.

01 — The story

I started at the surface and kept following the problem down.

I started where most do — apps, sites, backends, shipped for clients and app stores. Then my flagship project, Harrir, pulled me all the way down the stack: a search engine that understands Iraqi dialect, two AI agents with different jobs, a pipeline that retrains them from their own mistakes, and a GPU box that never sleeps. I built all of it — because production AI only works when one person can trace a bad answer from the customer's screen down to the KV cache.

Agents & orchestration

LangGraph graphs with deterministic safety nets, human-in-the-loop interrupts, and cross-agent handoffs a fine-tuned model can't break.

RAG & retrieval

Hybrid dense and sparse pipelines, bilingual BM25 with real Arabic normalization, rank fusion, reranking, and chunking that survives re-ingestion.

Evaluation & serving

Champion-vs-challenger replay gates with confidence intervals, LoRA hot-swaps with no restart, and the whole stack on one GPU I have to fit it into.

Product engineering

Next.js sites, React Native and Flutter apps on real app stores, Supabase backends — the product muscle that keeps the AI honest about users.

02 — Experience

Where the scars came from.

Propelius Technologies

Surat, Gujarat, India

Software Engineercurrent

Sep 2024 — Present · Full-time
  • Own two agentic AI systems in production, an AI stylist and a support agent, serving about 2.3K unique customers and 4.4K messages a month across Android, iOS and web.
  • Shipped AI product discovery handling roughly 70K requests a month across text and image search over a catalog of 11K products and 60K variants, replacing keyword search with semantic and visual retrieval.
  • Run the entire AI stack — chat, vision, embeddings and two fine-tuned adapters — on self-hosted GPU infrastructure at about $500 a month, with a costed path to $300 on reserved capacity, instead of a paid inference API.
  • Built the evaluation and release pipeline that has shipped 6 adapter versions through a champion-versus-challenger replay gate with 95% confidence intervals. One release raised language-consistency accuracy from 0.754 to 0.965 and tool-choice accuracy from 0.833 to 0.900 on held-out conversations.
  • Delivered features across 10+ client projects beyond the flagship platform, spanning landing pages, admin panels, web platforms and AI-driven systems on Next.js and Supabase with LangChain and LangGraph.
  • Set up and operate the AWS infrastructure behind these systems (EC2, S3, nginx) and work directly with clients through requirements and demos.

Blue Data Consulting

Surat, Gujarat, India

Intern

May 2023 — Mar 2024 · Internship
  • Built React interfaces and learned to containerize and ship applications with Docker.

03 — Education

The formal part.

Bachelor of TechnologyComputer / Software Engineering

Uka Tarsadia University

Sep 2021 — Aug 2025

Surat, Gujarat