Junior AI Engineer

About Humai

Humai builds AI you can vouch for. Founded in Dubai in 2025, we exist for one problem: companies have adopted AI but will not rely on it for work with consequences. Our answer is one accountability model across two products. Huscribe for what AI does: it checks agentic workflows against what their owner approved. Mairit for what AI produces: it puts a qualified, accountable name on the work.

Audited to SOC 2 Type II and certified to ISO/IEC 27001:2022, both products in scope. Salesforce partner. Small team, founder-led, no layers.

The role

You join the AI team behind Mairit and Huscribe and learn by shipping. You work directly with our senior AI engineer, our senior backend engineer and the founder on the pipelines behind both products. Within months, parts of those pipelines are yours.

We hire juniors for judgement and care, not years. Show us something you built and how you tested it.

On-site in Dubai. We sponsor the employment visa.

Why take this seat

• Production from month one. Pipelines with paying customers, and your code in them.
• People. Senior AI and backend engineers review your work every day, and the founder is in the room.
• Growth. A junior title we expect you to outgrow. We promote on what you own, not on tenure.
• Range. Frontier models, open models, evaluation, data, documents, MCP. You touch the whole stack in your first year, and you build with the best AI tools from day one.
• Pay. A salary that grows with what you take on, a UAE employment visa, and medical insurance.

What you'll do

• Build and maintain evaluation sets and graders for our pipelines, run experiments, and report what changed
• Implement pipeline steps under senior review: prompts, checks, parsers, retrieval
• Curate clean, anonymised corpora from reviewed work: terminology, exemplars, test cases
• Work on document processing: OCR output, layout, rendering checks
• Trace failures end to end, then fix them or escalate with a clear write-up
• Build small internal tools the ops and product teams use every day

What good looks like after a year

• You own the evaluation sets for at least one pipeline and the team trusts your numbers
• Steps you wrote run in production every day
• You can take a failed job, find the cause, and explain it in five sentences
• You have shipped something a colleague uses without being asked twice

What you bring

• A degree in computer science, engineering, maths or similar, or equivalent proof: projects, open source, internships
• Solid Python. Comfortable with Git, tests, HTTP APIs and SQL
• Hands-on LLM work: you have built at least one real thing (a RAG app, an agent, an eval harness) and can show it
• You understand why "it looked right" is not evidence: test sets, precision and recall, regressions
• You read other people's code without complaint and ask before guessing
• You write clearly

Nice to have

• Arabic reading ability
• LangGraph, LangChain or DSPy
• OCR or document AI
• Vector databases
• Azure

Not for you if

• You would rather wait for instructions than ask
• Details bore you
• You want to work only on the model and never on the data
• You want to work remote