AI engineer working the full lifecycle — from scraping raw data and fine-tuning LLMs to tracking every run in MLflow & W&B, containerising with Docker, and shipping it behind a production backend.
Getting a model out of the notebook and into something people can actually use.
I work across the whole machine-learning lifecycle — from cleaning messy, scraped data to fine-tuning language models and running them on constrained edge hardware like the Jetson Nano.
Lately I’ve been building agentic and RAG systems: I fine-tuned Qwen 1.5B on a hand-curated dataset so it answers domain questions in a specific voice, and I stand these models up behind FastAPI backends with scraping and browser automation feeding them real data.
The part I care about most is that the work is reproducible. Every experiment goes into MLflow or Weights & Biases, so a model can always be traced back to the exact data, parameters and code that produced it — and I quote honest validation numbers, not the flattering random-split ones. Environments are pinned with Conda and shipped in Docker, so what trains on my machine is what runs on the server.
I ship the product around the model too, not just the endpoint. On JDWIN that meant a passwordless auth flow — Twilio SMS one-time codes and Brevo transactional email codes, with Google and Apple sign-in beside them — so a token is only ever issued after an identity is genuinely proven.
I also teach — I like breaking hard AI concepts down until they click, which keeps my own fundamentals sharp.
Grouped by where they show up in the pipeline.
From a geospatial ML product running in production to LLM fine-tuning and computer vision — shipped systems first, deep-dive experiments after.
Got a model to fine-tune, a RAG system to stand up, or an idea that needs to leave the notebook and reach real users? I’m based in Dammam and happy to talk.