Senior AI Engineer – Reinforcement Learning Lead

Duku AI; Greater London; On-site; GBP 80,000 - 120,000; Full time

Job summary

A pioneering AI company in London seeks a Senior AI Engineer – Reinforcement Learning Lead to innovate in software quality. The role involves shipping ML to production, focusing on reinforcement learning, and developing autonomous agents. You'll work closely with experienced leaders, pushing boundaries in AI technology. Join a mission to redefine software testing while enjoying equity and substantial resources to aid your development in a fast-paced startup environment.

Benefits

Equity stake

Unlimited resources

Impactful work

Qualifications

• 5+ years shipping ML to production (real systems, not papers).
• Deep RL expertise, you think in Q-values and policy gradients.
• Experience building autonomous agents that actually work at scale.
• Python/PyTorch mastery.

Responsibilities

• You’ll see your reinforcement learning prototypes running live inside real applications.
• Agents will have evolved, scaling across multiple environments.
• The intelligence you’ve built will sit at the heart of every release.
• Working side by side with a CTO and an elite founding team.

Skills

Deep RL expertise

Python mastery

Experience building autonomous agents

5+ years shipping ML to production

Job description

Senior AI Engineer – Reinforcement Learning Lead

Change Software Forever

QA slows the world down. Flaky tests kill trust, stall releases, and bleed engineering velocity.

Duku AI is ending that era.

We’re building autonomous agents that think like engineers: they run every critical user journey, catch failures before users do, and self-heal as the codebase evolves. Real AI teammates, not test scripts that break on impact.

We’re venture-backed and led by operators who’ve scaled Meta’s testing infrastructure, launched Uber’s global playbooks, and grew Deliveroo from zero to hypergrowth. We know what elite execution looks like and we’re hunting for one more builder to help us rewrite the rules of software quality.

Why This Role is Different

Most “AI engineer” jobs are just applying models someone else built. This isn’t that. This is about pushing RL to its edge:

• Agents that think: networks that see and understand apps through vision, structure, and behavior.
• Agents that explore: curiosity-driven RL that uncovers edge cases no human would think of.
• Agents that learn: smarter with every bug, sharper with every correction.
• Agents that scale: millions of states, thousands of sessions, decisions in sub-seconds.

If you’ve ever wanted to take RL out of papers and into the wild, this is it.

What You’ll Achieve

In your first three months, you’ll see your reinforcement learning prototypes running live inside real applications, surfacing bugs no human ever noticed.

By six months, those agents will have evolved , scaling across multiple environments, learning and adapting in ways that prove this isn’t theory but reality.

And within a year, the intelligence you’ve built will sit at the heart of every release for our first customers, powering their ability to ship AI-generated code with confidence.

What You Bring ( Non‑Negotiables)

• 5+ years shipping ML to production (real systems, not papers).
• Deep RL expertise, you think in Q‑values and policy gradients.
• Experience building autonomous agents that actually work at scale.
• Python/PyTorch mastery.

The Stuff That Matters

• You’re obsessed with solving “impossible” problems.
• You’d rather ship and learn than debate in theory.
• You can explain RL to a CEO and optimize it for a GPU cluster.
• You thrive in chaos and see it as opportunity.

Why Join Now

• Impact: You won’t be “joining a team.” You’ll be the team that defines how software is built in the age of AI. Your code won’t sit in a corner , it will become the backbone of a new category.
• Market: Software testing hasn’t changed in 30 years. AI‑generated code has rewritten the rules overnight. Whoever solves this bottleneck doesn’t just win a market , they reshape the entire industry.
• Team: Small, elite, no passengers. You’ll be working side by side with a CTO who built this at Meta and a founding team that’s scaled some of the fastest‑growing tech companies on the planet.
• Timing: Rarely do technology shifts and career timing line up. This is one of those moments. Five years from now, autonomous QA will be a given. Right now, it’s unsolved , and you could be the one who solves it.

The Challenge

Big tech tried to brute‑force this problem and hit a wall. Most startups never got past brittle scripts. The reason is simple: building true autonomy takes more than patching frameworks , it takes intelligence. That’s the path we’re on. Your system will need to:

• Navigate the chaos of modern web apps.
• Learn from sparse, delayed rewards.
• Balance exploration with validation.
• Transfer knowledge across completely different applications.

It won’t be easy. That’s the point.

What You Get

• Equity that actually moves the needle , not token options, but a real ownership stake in what could be the category‑defining AI company of the decade.
• Unlimited firepower , the hardware, compute, and resources you need to push RL further than anyone has before.
• A seat at the table, not a cog in the machine, you’ll be in the room where every decision is made, shaping both the product and the company.
• Speed over politics , a London base where execution beats process, every time.
• A shot at legacy , work that will outlive your CV, the kind of achievement you’ll still be talking about 20 years from now.

To win the space, we’re looking for the best people in London, with 10/10 ambition and work ethic to join us and build a product people love.