Do work
worth naming.
Four open roles. Zero bench. Names you'll point to.Your code will reach ministries, energy majors and NASDAQ-listed clients — and power an AI product platform with thirty-five products on its roadmap. We hire few, brief hard, and answer every applicant. Read the roles like mission briefs, because that's what they are.
Not a typical
IT shop.
Most IT services careers pages promise culture. We'd rather show you receipts — and tell you plainly what working here is like.
You're hired for named work
Nobody sits on a bench waiting for a project. Every role here maps to real engagements and real products — even embedded roles carry names like ADNOC-Fertiglobe and federal ministries.
◆ True for every role on this pageThe proof is public
Seventeen named case studies, six countries, government to NASDAQ-listed clients. Read them yourself — then decide if you want your work on that page.
◆ ADNOC · Amazon · SAP · SERCO · MinistriesAI in your hands, day one
Copilots, LLM APIs, eval frameworks — the modern AI toolchain is standard issue here, not a request form. We ship with AI in the loop; you will too.
◆ Copilots · LLM APIs · Evals · AgentsServices and product, both
We're not only a services firm — we're building MoxSuite: an AI-native platform with 35 products on the roadmap. Services sharpen the product; the product sharpens you.
◆ 6 products in motion · 35 on the roadmapYour work travels far
No layers of hierarchy between your commit and a client in Riyadh, Abu Dhabi or California. Ownership is real because the team is lean — and so is the accountability.
◆ Commit → client. No layers between.No ghost pipelines
You apply, you hear back — acknowledged within 48 hours, decided in days, not months. We treat candidates the way we treat clients, because some of you will become both.
◆ 48-hour acknowledgement, in writingFour briefs.
Find yours.
Each role lists what you'll ship in your first ninety days and the bar we screen against. If the brief reads like your last two years — apply.
Build the intelligence core of MoxSuite and ship enterprise AI features that named clients actually use — not demo-ware.
- Ship an LLM-powered feature into MoxSuite production
- Harden our bilingual EN/AR assistant patterns — Mattex-grade — into reusable modules
- Stand up evals, guardrails and telemetry for a client-facing AI workflow
- You've shipped LLM applications to real users — RAG, agents or assistants in production
- Strong Python plus one production stack; Node/TypeScript welcome
- You treat AI like software: evals, guardrails, observability — not vibes
- EN/AR content sensitivity is a plus; GCC market context is a real plus
Own AI architecture from enterprise pre-sales through production — the technical spine ministries and energy majors trust in the room.
- Author two winning solution architectures for GCC enterprise AI engagements
- Define our reference architecture for agentic workflows across the practice
- Stand beside BD in ADNOC-grade rooms and carry the technical conversation
- Enterprise solutioning scars — you've scoped it, defended it, and then delivered it
- Deep GenAI patterns: RAG vs fine-tuning trade-offs, agent orchestration, cost mathematics
- Governance fluency: data residency, PDPL-class constraints, security reviews
- Client-facing gravitas; GCC exposure strongly preferred
Design the data foundations everything else stands on — enterprise client estates and the MoxSuite shared data fabric.
- Blueprint version one of the MoxSuite shared data fabric
- Design an AI-ready analytics estate for an enterprise client — UTAC-class scale
- Set our modelling standards so every new pipeline ships faster than the last
- You've owned enterprise data architecture end-to-end, not just a corner of it
- Dimensional and lakehouse modelling both; governance instincts built in
- Power BI estates at scale; SQL that reads like prose
- You design for the AI queries coming next year, not just today's reports
Build the pipelines and automation that quietly remove hours from client payrolls — White City-grade, where ~80% of an operation runs itself.
- Own a production pipeline behind a live client automation
- Instrument telemetry so operators see revenue daily, not monthly
- Automate one manual workflow end-to-end — and prove the hours it saved
- Production Python and SQL — you've been paged at 2am and fixed it
- Orchestration experience: Airflow-class ETL/ELT in the real world
- APIs, webhooks, integration glue — you make reluctant systems talk
- A bias to measure: if it saved hours, you can show the number
Four steps.
Days, not months.
Apply
Send the form below with your portfolio, GitHub or LinkedIn — proof of work beats prose. Skip the cover-letter theatre; the "proudest build" question is the only essay we want. Every application is read by an engineer, not parsed by a keyword bot.
◆ Acknowledged in writing within 48 hoursConversation
Thirty minutes with a working engineer or architect — not a recruiter reading your CV back to you. We talk about what you've actually built, the trade-offs you made, and what you'd do differently. You should be interviewing us just as hard.
◆ Real technical talk, both directionsBuild Round
A practical, tightly-scoped exercise drawn from the kind of work you'd actually do here — never a free-labour project. We review it together like a real code review: your reasoning matters as much as the result.
◆ Your craft, visible — your time, respectedDecision
A clear yes or no, with reasons either way — because a good "no" still costs you nothing but earns our respect. If it's yes, the offer reflects the brief you just read, and your first-90-days plan is already written.
◆ No ghosting. Ever. That's the policy.Clear
the bar.
Pick your role, drop your proof of work, and tell us the thing you've built that you're proudest of. That last part matters more than the CV.
Apply now
If you clicked "Apply for this role" above, we've already ticked it for you. Multiple roles? Tick more than one — we'd rather see range than guess it.