UROOJ
FATIMA

I work across the interface, the workflow and the model underneath.
Neu is my AI product design and development practice. I work with teams who need an agent to handle a task, a workflow to connect their tools or a model trained on examples from their domain.
I am interested in the decisions that give these systems their shape. What can an agent act on? What should someone be able to undo? Which examples teach a model the behaviour a product needs?
My work sits between design and engineering. I sketch the review screen and write the code behind it in the same week, because the two depend on each other. A model that is right most of the time is only useful if the interface makes the remaining cases easy to catch.
Four rules I work by.
Show the evidence
Every proposed action carries the sources it used.
Make it undoable
Automate what is easy to reverse. Ask a person for the rest.
Measure first
A baseline comes before any claim of improvement.
Hand it over
Your team should be able to change it without me.
Where I spend the hours.
- Interface
- Review screens, approval flows and the moments where a person and a model share a task.
- Workflow
- The rules that move work between tools, and the exceptions that should stop it.
- Model
- Prompts, evaluation sets and tuned adapters, measured against a fair baseline.
- Handover
- Documentation, tests and a walkthrough so your team owns what was built.
What you can commission.
- 01Agentic systemsI build AI agents that can look up information, choose which tool to use next, and hand a decision to someone on your team.4 to 8 weeks
- 02Workflow automationI connect the forms, inboxes and apps your team already uses so routine work moves between them automatically. People see the exceptions that need a decision.2 to 6 weeks
- 03Model tuningI adapt a language model to your categories, terminology or output format, then compare it with the original on examples it has never seen.3 to 6 weeks