AI & data science
AI Engineer
Build and evaluate the document pipelines behind reports a clinician or an accountant signs, grounded in the client's own records.
- Location
- Lahore, Pakistan
- Type
- Full-time
- Experience
- 2 to 4 years
- Posted
- 4 August 2026
What you will do
- Build retrieval and generation pipelines over a client's own documents, tickets and ledgers.
- Design the evaluation harness before the model: measure a draft against the corrections a reviewer actually made.
- Structure output to the format a department already writes in, rather than asking it to change.
- Instrument every draft so the citation, the query and the pre-edit version can be inspected later.
- Sit with the reviewing professional during evaluation and watch what they change.
What we are looking for
- Two to four years writing production Python, including at least one LLM or ML system that real users depended on.
- Retrieval systems in depth: chunking, embeddings, reranking, and the failure mode of each.
- SQL and enough data modelling to know what a set of records can and cannot support.
- Willingness to learn a domain you do not know well enough to ask the expert a useful question.
Nice to have
- Worked with clinical, financial or statutory documents.
- On-device or edge inference.
- A view on when fine-tuning is not worth it.
What happens next
- 01
Apply
Send the form with a CV and anything you have built. A person reads every one.
Reply within 5 working days, either way
- 02
Intro call
Thirty minutes with the person you would work for. What you have built, what you want next.
30 minutes, remote
- 03
Technical or portfolio conversation
Ninety minutes on your own work, or a real problem from our backlog. No whiteboard trivia.
90 minutes, remote or in Lahore
- 04
Offer
A written offer with the level, the scope and the number in it. Questions answered before you sign.
Within a week of the last conversation