AI/LLM engineering
Production LLM systems: multi-agent architectures, RAG pipelines, Claude and OpenAI integrations, evals, prompt engineering at scale. Shipped in my own products, built for clients.
Work with me
the one page to read if you're deciding whether to hire me
Who this is for
Founders and small teams who need production AI systems built, or rescued. Companies that shipped fast (with a tool, an agency, or a hurried team) and now need the codebase to survive real users. Anyone who wants a senior technical partner without a full-time hire.
The common thread: I've been the sole technical owner on every engagement for years. You get someone who takes the whole problem, not a ticket-taker.
What I do
Vibe-code cleanup & pre-launch review
Someone (or some tool) shipped you a codebase that "works" but isn't production-ready. I audit what's there and hand you a written report (what breaks, what's missing, what order to fix it in) with the highest-risk fixes applied.
Full-stack AI product development
End-to-end builds: backends in Go, Node, or Python; AWS infrastructure; PostgreSQL; async pipelines; the AI layer integrated properly. "It scales" and "it doesn't fall over at 3am" both non-negotiable.
Fractional CTO / technical advisor
A senior technical partner without the full-time hire. Architecture reviews, technical roadmap, code review, due diligence on existing systems. Strongest fit: early-stage AI products.
Proof, not promises
- LegalSort.ai: an AI legal document classifier on multi-agent LLM pipelines. Built solo, end to end, shipped.
- 3+ years as sole technical founder of Due, a legal-tech SaaS: 750+ users, 50K+ documents, every technical decision from architecture to production incidents.
- Upwork: 40+ projects, 4.9/5 ⭐️, Top-Rated. Client relationships measured in years, not gigs.
- Kilanka: a $17.5k hybrid app (TypeScript, React, Capacitor), one codebase shipped to web, iOS, and Android. 390 hours end to end, 5.0 ⭐️.
- Open source: the Gas backend framework, the foundation under LegalSort and this site.
Kind words (from real clients)
"Ahmed went above and beyond the call of duty, bringing his own ideas into the mix and giving us a better product than we ever had before. He was very responsive to our questions and went out of his way to meet with us for long periods of time. Truly a wonderful person that has been a fantastic person to work with."
"Ahmed is excellent. We have worked together for over 3/4 years and would highly recommend him to other businesses. This task was highly critical to one of our clients and Ahmed performed very well, was always available and I could trust him to liaise directly with our client."
"I cannot speak my praises of Ahmed enough. He has provided us with high-quality work in a timely manner. Not only that, but he has gone out of his way to overhaul the UI in our app and provided us with a new perspective. He knows how to work under stress and timelines and has demonstrated excellent resilience and adaptability while helping our startup through this time."
"Great reliable full-stack dev. His experience with multiple programming languages was a great resource as we navigated different tech options. He provided invaluable guidance as we were designing the product and has continued giving creative and state-of-the-art advice for both back-end and front-end solutions. Definitely recommend."
How it works
Three shapes, pick what fits: a scoped project (defined outcome, defined price), a retainer (ongoing build or maintenance, weekly cadence), or advisory (architecture reviews, code review, roadmap: hours, not weeks). Not sure which? The pre-launch review above is the usual first engagement: fixed price, fixed scope, and you keep the report either way.
The first conversation is a free 15-minute call. Tell me the problem; I'll tell you how I'd attack it.