Senior Data Scientist
- Vollzeit
Position: Senior Data Scientist – AI Engineering
Department: Logistics Tech – Data Science & AI
Reporting to: Data Science Manager
Location: Amsterdam, Berlin, London
Hybrid role: 3 days a week from our office & 2 days working from home
Full/Part time: Full time
Ready for a challenge?
Whether it's a Friday-night feast, a post-gym poke bowl, or grabbing some groceries, our tech platform connects tens of millions of customers with hundreds of thousands of restaurant, grocery and convenience partners across the globe.
About this role
As a Senior Data Scientist – AI Engineering, you will shape how we build, evaluate, and operate the AI agents that support the people who keep our delivery network moving. These are production, user-facing agents that help users resolve issues and take actions in real time, across multiple markets and languages.
Your mission is to make these agents measurably better and safe to change — and to make that the way the whole team works. You will influence the technical direction of agent quality: how we decide that an agent is good enough to ship, how we know when it stops being good in production, and how we release changes with confidence. You will take ambiguous problems from idea to production with little supervision, and raise the bar for the team through design and code reviews. You will partner with Product, Engineering and Operations, and help us make sound architectural choices as agent technology evolves quickly.
These are some of the key components to the position:
- Agent Architecture & Technical Direction: Lead the design of agent systems — agent and workflow architecture, tool integration, and human handover — and make well-reasoned trade-offs between quality, latency, cost and maintainability. Evaluate new models, frameworks and agent patterns, and recommend when to adopt them and when not to.
- Evaluation Strategy: Define how we evaluate agents end to end: offline evaluation datasets, LLM-as-a-judge and deterministic metrics, automated end-to-end testing, and online evaluation that links agent quality to business outcomes. Establish the standards and shared tooling that let the whole team make reliable ship / don't-ship decisions.
- Observability & Reliability: Own the approach to monitoring agents in production: tracing, quality, cost and latency signals, and alerting, and a clear process for investigating regressions and incidents. Turn production traces and user feedback into a prioritised improvement loop.
- Safe Delivery: Define how agent changes are released and rolled back — feature management, staged rollouts, and guardrails — and drive the engineering practices (testing, code review, CI/CD) that make this routine rather than risky. Make sure data privacy, security and compliance are built in.
- Technical Planning & Delivery: Translate product requirements into technical plans, break down complex work, flag risks, and set realistic timelines. Identify and prioritise technical debt and platform investments alongside feature work.
- Impact & Insights: Quantify the business impact of agent improvements using sound experimentation and causal reasoning, and communicate clearly to senior stakeholders, including trade-offs and uncertainty.
- Mentoring & Team Practices: Guide Data Scientists through code and design reviews, pairing and feedback, and help shape hiring, onboarding and team practices.
- Community & Influence: Collaborate with Backend, Operations and Product teams, and help grow an internal community of people building user-facing agents at JET — sharing patterns, lessons learned and reusable tooling.
What will you bring to the team?
- Substantial experience designing, building and operating user-facing AI agents in production, including having lived through real-world failures, regressions and incidents, and the lessons that came from them.
- Deep experience with agent evaluation: designing evaluation strategy and datasets, building and calibrating LLM-as-a-judge metrics, automating regression testing, and connecting offline results to online outcomes through well-designed experiments.
- Deep experience with agent observability and monitoring: designing tracing and quality-monitoring for multi-step agents, defining metrics and alerts, and using production data to drive improvements (e.g. OpenTelemetry, Datadog, or similar tooling).
- Strong hands-on experience with agent frameworks (e.g. Google ADK, LangGraph, or similar), tool use / MCP, and multi-step or multi-agent architectures, with the judgement to know when added complexity is worth it.
- Experience defining or operating release processes for production ML or agent systems — feature management, staged rollouts, and rollback.
- Excellent Python and SQL, and strong software engineering practice — typing, testing, CI/CD, code review, and designing maintainable systems that others can build on.
- A strong data science foundation: statistics, experiment design and causal inference, applied to ambiguous, real-world problems.
- A track record of leading technical work across teams, taking projects from idea to production, and mentoring more junior colleagues.
- Strong communication skills — able to explain complex trade-offs to technical and non-technical audiences, and to influence without formal authority.
- Nice to have: experience designing agent evaluation strategy at org scale, containerised deployments (Docker/Kubernetes), a basic understanding of Kotlin, the language our Backend Engineers use.
At JET, this is how we play
Our teams forge connections internally and work with some of the best-known brands on the planet, giving us truly international impact in a dynamic environment.
Being the best at what we do isn't just about delivering on our strategy. It's a competition for something incredibly valuable – our customers' choice. Every time a customer decides where to order, they're picking a side.
At the heart of the JET Customer League are our values and behaviours. They guide every interaction, every decision, every innovation. These are the actions we need to perform consistently and brilliantly, to surpass the competition and earn our customers' loyalty, again and again.
Fun, fast-paced and supportive, the JET culture is about movement, growth, helping one another to succeed and celebrating wins. By truly living our values and embodying our behaviours, we're building a customer-first culture which enables us to stay one step ahead of the competition.