Toufiq Qureshi — Full Stack AI Engineer

about

Backend to UI, I own the whole stack.

I'm a Full Stack AI Engineer based in Vasai, Mumbai — self-taught, no formal degree, two years of hands-on production experience. My work sits at the intersection of backend engineering and agentic AI: designing APIs and data models, then building the AI layer that sits on top of them and actually solves operational problems.

I'm comfortable making and defending architecture decisions — choosing FastAPI over Django for its async performance, designing caching layers to cut latency and cost, and redesigning AI systems when they're burning tokens on the wrong problem. I care about systems that run in production under real constraints, not ones that only work in a demo.

Currently working full-time at Revmerito as a Full Stack AI Engineer (official title: Marketing Data Scientist), while building Staybooker and contributing to open-source AI tooling on the side.

system.info

locationVasai, Mumbai

focusFastAPI · React · Agentic AI

experience~2 years

educationSelf-taught

statusopen to work

experience

Where I've worked

OCT 2024 — PRESENT

Full Stack AI Engineer

Revmerito · Mumbai, India (official title: Marketing Data Scientist)

Architected an agentic automation framework (CrewAI, LangChain, SeleniumBase) for distributed web scraping — processing 800–1000 URLs daily, eliminating 150+ manual hours a month
Engineered a real-time SEO analytics chatbot integrating Search Console and Analytics APIs with LLM-powered anomaly detection
Built a multi-agent keyword research system (research → clustering → competitive analysis → forecasting), cutting campaign ideation from 5 days to 15 minutes
Built an MCP server (FastMCP) connecting Claude Desktop to the company SQL database — letting a 10-person ops team query multi-hotel data in natural language instead of writing SQL

selected work

Things I've built

Staybooker

actively maintained

A solo-built, production-grade multi-tenant hotel booking SaaS. Router-based multi-agent AI layer, Razorpay + WhatsApp webhook pipelines tested end-to-end with live test transactions, Redis caching with automatic in-memory fallback, and strict per-hotel data isolation.

195+

API endpoints

70%

LLM cost cut

570+

passing tests

74K

lines of code

FastAPIReactSupabaseRedisAgnoRazorpay

Google Reviews AI Manager

Lamatic AgentKit · PR open

A real, end-to-end Kit contributed to Lamatic's open-source AgentKit as part of their Applied AI Engineer hiring challenge. Real Google My Business API integration with multi-location support, a Lamatic AI flow for contextual review replies, and edit-before-publish UI that posts live back to Google.

Next.js 15NextAuthGoogle My Business APILamatic AI

Agentic Content Generation Platform

shipped

A five-agent pipeline (Requirement → Content → Quality Check → SEO → Publishing) that fully automates content production and auto-publishes to a live CMS. Ollama for local inference to keep per-article cost at $0.

70%

faster production

OllamaAgnoSearXNGDocker

AI Candidate Analyzer

Agno Global Hackathon

Evaluates candidates against a job description using their real GitHub activity and LinkedIn trajectory — not just a self-reported resume — to help HR managers shortlist faster.

AgnoGitHub APIExa SearchStreamlit

stack