Ali Sajid
I take AI products from prototype to production — and build the full stack underneath them: retrieval, models, API, interface, pipeline.
Ten years, ten roles, three countries — Pakistan, then the UAE, now Germany, with US and European clients served remotely. Fintech, healthcare, enterprise, government. Lately: enterprise LLM systems for the Department of Culture and Tourism in Abu Dhabi, and full-stack AI products at ai71.
Now in Berlin. Open to a senior role, or to contract work where a team has the demo and needs the system.
Ten years. Ten roles. Three countries. Five enterprise AI tools shipped in a single year.
Two ways in. Both start with an email, and I reply personally.
A senior engineer who owns the whole system.
Ten years, ten roles, three countries — fintech, healthcare, government, shipping, property analytics. Lately enterprise LLM systems for the Department of Culture and Tourism in Abu Dhabi, and full-stack AI products at ai71.
What I own
Architecture through deployment — schema, API, interface, CI/CD. Most recently five AI applications from concept to production on Next.js, FastAPI, PostgreSQL, vector databases and Azure DevOps.
How I lead
Squads of four to seven, onsite and offshore. Architecture, breakdown, code review, delivery on stated dates. Interviews and onboarding where a team needs that too.
What I am not
An AI specialist who needs someone else to build the product around the model. The end-to-end engineering is the point; the models are one layer of it.
Berlin, Germany · Immediate joiner.
Get it out of the prototype and into production.
Contract and consulting work, on the systems below. I design it, build it, and hand over something your own team can run.
Prototype to production
A demo that works on a good day, turned into a system that holds: retrieval over your own data, scoring that shows its reasoning, and a deployment pipeline someone else can run.
Internal tools over your own data
Semantic search across records that were never searchable. Document and tender analysis. A conversational layer over the dashboards people already have. Most recently on SAP HR data, for a government department.
A product, end to end
Schema, API, interface, deployment, tests — one pair of hands across the whole thing. Loan processing to PSD2 and GDPR; telemedicine; property analytics. Plus technical leadership for a small team, where that is what is missing.
Berlin or remote. Direct reply.
Recruiters ask in plain language. The system answers from the whole application history.
An internal recruitment intelligence platform for DCT Abu Dhabi, built on SAP HR data: semantic search over historical candidates, LLM scoring against the role with its reasoning shown, and shared annotation so a panel evaluates together instead of in sequence.
Four more followed in the same year — job-description generation, procurement and tender analysis, an enterprise knowledge assistant, and a conversational layer over Power BI.
All projects →Trade advisor, novel reader
FastAPI services with layered routing, business logic and LLM integration — one for Pakistan's export data, one that maps characters across Project Gutenberg books.
CSV RAG, ATS fit analyzer
Question-answering over tabular data in NestJS; CV-to-job scoring with stated reasoning on Next.js and Groq.
Fintech, health, real estate
White-label loan processing to PSD2 and GDPR; telemedicine monitoring between hospital visits; US property investment analytics.
AI & LLM
Large language models · Generative AI · Prompt engineering · Embeddings and semantic search · Retrieval-augmented generation
Backend
Node.js · TypeScript · Python · FastAPI · NestJS · Express · GraphQL · Microservices · WebSocket · RabbitMQ
Frontend
React · Next.js · TypeScript · Tailwind CSS · Chrome extensions
Data & cloud
PostgreSQL · MongoDB · Vector databases · Azure · AWS · GCP · Firebase · Docker · CI/CD
Leading
Squads of four to seven · Systems design · Code review · Scrum · QA automation with Cypress and Jest
“The kind of engineer every team needs. Not just a full-stack engineer: a problem solver.”
I started in games — Unity, C#, a wolf simulator on the Play Store — which is still how I think about software: build the thing, let someone play with it, watch what breaks.
I like the unglamorous parts. Schema design. Naming. Deleting code. A build that goes green without anyone holding their breath.
Learning German, slowly and stubbornly. Follow the animal-welfare and education work when I have hours to give. Happy to be the person who asks the obvious question in a meeting.
Tell me what's stuck.
Hiring, or need a system finished — either way, email is fastest. I answer the same day.
Colophon. Set in Source Serif 4. Photographs printed as four misregistered process plates.
© 2026 Ali Sajid. Berlin, Germany.