Forward Deployed Engineer (Dubai)

Xanadu Technology Limited; Dubai; On-site; AED 420,000 - 640,000; Full time

Benefits offered by this job

Equity (ESOP)

Job summary

Xanadu Technology Limited seeks a senior AI/ML lead who will sit with customer leadership and frontline teams to map AI value, design reference architectures, and drive production deployments. You’ll prototype quickly, then guide a pod of engineers to deliver real-user solutions.

You will work closely with Canva and Anthropic in-region, with a focus on adoption, metrics, and ongoing optimization. The role blends consultancy with engineering leadership, requiring clear communication, robust

Qualifications

• 6+ years in software or technical roles with customer-facing experience.
• Led small engineering teams or tech leads on delivery.
• Shipped LLM-powered software into production with prompts, agents and tool use.

Responsibilities

• Design reference architecture for customers’ environments with security and data handling.
• Prototype initial solution in the first weeks and hand over to the pod for production.
• Lead a pod to deliver production-grade agents with real users and post-go-live support.
• Shape proposals and Statements of Work based on client engagements.
• Identify reusable patterns to improve the Agent Library.

Skills

Customer-facing leadership

LLM-powered software

Python/TypeScript coding

Workshop facilitation

Onsite client engagement

Tools

Claude

Agent SDK

MCP servers

Python

TypeScript

Job description

About Xanadu

Xanadu helps mid-market organisations in the Middle East move from AI experiments to AI that actually runs in production. We're a Certified Services Partner, Claude Partner Network, building on Claude for the reasoning, workflow and agent layer, and a Canva partner for the creative and communications layer. Prompt to presentation.

We're Dubai-based, small on purpose, and in the room with our customers. We're building our AI practice right now, which means the people who join in the next six months get to shape how it runs.

The role

You're the person in the room. You sit with a customer's leadership and their frontline teams, work out where AI will genuinely move a number, then make it happen: design it, build the first version yourself, and lead a small engineering pod to get it into production with real users.

Half consultant, half engineer, fully accountable for the outcome.

Most FDE roles at the big labs sit with a sliver of a giant enterprise. Ours sit with the whole of a 500 to 5,000 person business. You'll see the CEO's problem and the finance team's problem in the same week, and fix both.

What you'll own

• Discovery. Run the workshops and interviews, understand the customer's existing architecture and integrations, map how work really flows, and turn "we want to use AI" into a scoped, priced piece of work with a number attached.
• Design. A reference architecture for the customer's environment, integrated with their systems of record, with security and data handling their risk team will sign off.
• Build. Prototype it yourself in the first weeks (Claude, Claude Code, Agent SDK, MCP servers, sub-agents, skills), then hand the volume to your pod and review what they ship.
• Production. Agents live with real users, not demos. You give the final sign-off before anything goes live. Evals, monitoring, runbooks, handover and 30 days of hypercare.
• Adoption. Stay close after go-live. Measure usage and impact, fix what isn't landing, find the next function. Success metrics are set in each Statement of Work, and you're accountable for hitting them.
• Reuse. Every engagement makes our Agent Library better. If you built it twice, it should have been a pattern the first time.
• Commercial. You're first to spot the next piece of work, and you help shape proposals and Statements of Work.

How you'll work

• You lead a pod: you onsite, and in our build hub a Senior AI Engineer, an AI Engineer and an AI Data Engineer, with the Agent Library behind you.
• You work alongside the field teams at Canva and Anthropic in-region. Partners first: we make them look good.
• Async inside the business, in person with customers.

What good looks like

• 90 days: Claude-certified, one discovery led end to end, first agent in production or a clear path to it inside a signed engagement.
• 12 months: two to three customers with agents running in production, a case study you'd be proud to put your name to, and at least three patterns you built now living in the Agent Library.

Must-haves

• 6+ years in software or technical roles, with 3+ of those customer-facing: FDE, solutions engineer or architect, technical consultant, or you founded something technical.
• You've led a small engineering team, or been the tech lead on a delivery. You'll be directing engineers in another country, so you've done it before.
• You've shipped LLM-powered software into production: prompting, agents and tool use, evals, and what happens after launch.
• You can code. Python or TypeScript, well enough to prototype fast and review a pull request credibly.
• You can run a room: exec workshops, frontline discovery, and the awkward conversation when scope drifts.
• Messy problem in, clear plan with trade-offs out.
• High EQ. You make customers and colleagues better, and you don't need to be the smartest person in the room to get the outcome.
• Based in or relocating to Dubai, and happy to be onsite with customers most weeks.

Nice-to-haves

• Arabic.
• Real estate, financial services, retail, or QSR and franchise experience.
• Integration experience with Microsoft 365 and Azure, Salesforce, HubSpot or ERPs (where most mid-market data lives).
• Canva Connect APIs, or anything that closes the loop from prompt to presentation.
• Change management or enablement chops. You know adoption is the hard bit.

What this isn't

• Not a slideware strategy role. If you can't build it, it isn't the job.
• Not a heads-down engineering role either. If you'd rather not talk to customers, look at our build hub roles.
• Not a big-firm role with a bench and a methodology team. You'll help write the methodology.

How we hire

1. Intro call (30 min): motivations, what you've built, what you want.
2. Values and behaviours (45 to 60 min): how you work with people.
3. Bar-raiser (60 min): long-term potential and whether you lift the team.
4. Working session (90 min): we hand you a realistic customer problem. You break it down, scope it and sketch the build live, laptop open, Claude open.
5. Decision: you'll hear from us within 48 hours of every stage.

Why Xanadu

• Build the practice, don't inherit it.
• Senior onsite, engineering capacity behind you. You're not carrying every line of code.
• Mid-market customers, where one agent can change how a whole company works.
• Equity (ESOP), so you share in what you build.
• No egos, no passengers. We measure outcomes, not hours.