Ahmedabad-based engineer turning ideas into reliable software — AI that holds up in production at Fibre2Fashion, and products founders can launch in weeks.
Real users, in production. Tap any card for the full case study.
An automated newsroom agent that reproduces an editorial team's morning routine — harvests ~65 primary and ~75 secondary curated source pages across 17 countries for textile/apparel news, scores and de-dupes the best stories with an LLM, confirms each against its primary source, then drafts publish-ready articles. Two-phase with a human in the loop, streamed live over SSE.
A browser-first catalogue of free online tools — calculators, text utilities, developer helpers, converters, and AI-assisted workflows — built to be fast, searchable, and cheap to run.
A production-ready D2C storefront for the Indian market — catalogue, variants, cart, Razorpay checkout, accounts and a nine-screen admin, built to be branded and launched rather than built again. Nothing is tied to one product category.
Clear scopes. Fixed prices. AI baked in where it actually moves the needle — not as decoration.
RAG over your own content, agent workflows, and document pipelines — with the structured output, retries, token budgets, and observability that keep an LLM feature dependable in production.
Next.js and Supabase with auth, payments, transactional email, and an admin panel from day one. Scoped tight, launched in weeks.
Razorpay/Stripe checkout, subscriptions, webhooks. Plus OTP, transactional email, media, anything your product needs.
Core Web Vitals in the green, schema-marked content, structured data. Rank on the queries that move revenue.
The short version — for anyone Googling the name.
Manuj Rai is a full-stack & AI engineer based in Ahmedabad, India. At Fibre2Fashion he has shipped two internal production AI systems end to end — a human-in-the-loop news discovery and drafting agent, and a document-to-article pipeline — both built in Python and FastAPI with multi-provider LLM orchestration, and integrated into the company's existing ASP.NET admin portal. His backend foundation is ASP.NET Core, C#, and SQL Server. Independently he ships modern web products with Next.js, Supabase, Resend, and Razorpay / Stripe.
Recent freelance work includes Karloo free online tools, the Trendverse D2C ecommerce build and the Hair Crezz salon booking template. Manuj works remotely with clients worldwide from Ahmedabad, India (IST), with same-day overlap for Indian teams and comfortable hours for US / EU. A few candid shots from work and travel live in the photo gallery.
To hire Manuj Rai for a freelance brief or a full-time role, the fastest path is the contact page — every message gets a personal reply, usually within 24 hours. Hiring managers can grab the latest résumé, and he's also on GitHub, LinkedIn, and Instagram.
The things recruiters and clients ask first — straight answers, no forms required.
Yes — I'm open to full-time AI engineering, backend, and full-stack roles at startups and product companies, alongside select freelance projects.
AI and backend: Python, FastAPI, and multi-provider LLM orchestration (OpenAI, Claude, Gemini) with structured JSON output, token budgets, and Prometheus metrics — plus RAG and pgvector, and ASP.NET Core, C#, and SQL Server as the enterprise foundation. Automation: Playwright, Selenium, Pandas. Freelance: Next.js, React, Supabase, Razorpay/Stripe, Resend.
I'm based in Ahmedabad, India. I work fully remote for freelance clients and open to remote-first or hybrid full-time roles globally.
Project-based. A typical SaaS MVP runs $4k–$10k, an AI integration $2k–$6k. Larger builds are quoted after a free 30-minute discovery call.
Internal systems at Fibre2Fashion, built end to end and in daily editorial use: News Desk, a human-in-the-loop story-discovery and drafting agent, and AI Article Gen, a document-to-publish pipeline. Both are Python and FastAPI with multi-provider LLM orchestration, token budgets, structured JSON output and Prometheus metrics, integrated into the company's existing ASP.NET portal. This site's AI twin is the one you can try yourself — RAG over my own published content with pgvector and an HNSW index.
A typical SaaS MVP goes from kickoff to live URL in 3–6 weeks. AI integrations land in 1–3 weeks depending on data and surface area.
I read every message. Average reply time is under 24 hours — for full-time roles, freelance briefs, or just to say hi.