Applied Scientist

Applied Scientist – Machine Learning (Reinforcement Learning) | London (Hybrid) £110k

If applying reinforcement learning to real physical systems excites you — not toy problems, not simulations, but live operational environments — this is a standout role.

A fast‑growing AI company is looking for an Applied Scientist to design, train and harden RL agents end‑to‑end: from problem formulation and reward design through to federated deployment and on‑site inference. You’ll work at the intersection of ML, physics and engineering , reasoning about thermodynamics and equipment behaviour just as much as architectures and training dynamics.

What you’ll be doing

• Design + train RL agents for real‑world control
• Turn messy telemetry into ML‑ready problems
• Validate behaviour against physical principles
• Productionise models — federated training, on‑site inference, monitoring
• Support research + academic work

What you bring

• Engineering/physics degree
• Strong RL experience (deep RL, debugging, non‑trivial problems)
• Python + modern ML stack (PyTorch/JAX, NumPy, RL libs)
• Comfortable with time‑series sensor data
• Ability to turn ambiguous operational challenges into tractable ML problems
• Happy switching between research and practical engineering

Nice to have

• Classical control, MPC, HVAC, thermodynamics, power systems
• Simulation, digital twins, surrogate models
• GNNs, meta‑learning, offline/safe RL
• Federated learning, distributed training, edge ML
• Publications or open‑source work
• Sustainability‑focused optimisation experience

Why it’s exciting

You’ll help shape how AI interacts with the physical world , working on systems with real sustainability impact at global scale — and collaborating with experts across ML, engineering and infrastructure to deploy physical‑AI responsibly and reliably.

Contact me directly - james@dmcgglobal.com or call 07464 475 407 to find out more.