You build and deploy models used directly in electricity trading. You will take ideas from research papers through implementation, large scale experimentation, and deployment into production systems that operate under real market constraints. You will work with large, high frequency time series data and ensure models are reproducible, testable, and monitorable.

Tasks

• Develop and productionise ML models used for forecasting, optimisation, and decision support
• Design and run ML experiments, training pipelines, and backtests
• Collaborate with Data Engineers on data pipelines and feature generation
• Ensure models are reproducible, testable, and monitorable in production
• Write clean, maintainable, well tested production code

Requirements

Core skills

• Master's or PhD in a quantitative field (ML, statistics, physics, applied maths, engineering)
• 3 or more years of experience as an ML Engineer, or Engineer/Researcher in a quantitative role
• Proven ability to write clean production quality code
• Strong experience with Git and Docker in ML workflows
• Hands on experience running ML experiments, training, and backtests end to end
• Comfortable with mathematical optimisation and training ML or DL models
• Professional English (C1) is required

Nice to have

• Academic or industrial research experience
• Experience with deep learning, linear programming, stochastic optimisation, reinforcement learning, or time series analysis
• Experience working with meteorological or weather data
• Familiarity with trading or financial data, electricity markets, or power systems

Benefits

• You move in flat hierarchies and have short decision making paths
• End to end ownership from modelling through deployment
• Close collaboration with a small senior team and fast iteration cycles
• Flexible hybrid setup depending on location
• Compensation is based on the IT collective agreement (38.5 hours/week), with willingness to overpay depending on experience and qualifications

Electricity markets are entering a new era. As renewables, storage, and flexible demand scale, price dynamics get sharper, faster, and more volatile. Winning is no longer about reacting well, it is about making high quality decisions continuously. marbl is building the algorithmic flexibility trading layer for this market reality. We turn market data into forecasts and optimisation driven actions that can run live, so trading teams can scale short term decision making across portfolios without scaling headcount or operational risk. The outcome we are driving is simple: make flexible assets easy to trade well. Better decisions, better execution, stronger economics, and faster progress toward a low carbon power system.