[ ABOUT THE ROLE ]
Levent's Data Science team keeps models in production, not notebooks. We design, train, and operate Bayesian MMM, SKU demand forecasting, and multi-agent decision systems for enterprise clients across retail, FMCG, and financial services. You will own the full model lifecycle — from research and experimentation to live inference in client environments.
[ RESPONSIBILITIES ]
- Design and train forecasting and classification models for client deployments
- Build retraining pipelines that keep models current without manual intervention
- Contribute to multi-agent decision loops — scoring, routing, fallback logic
- Write production-grade Python code with full test coverage
- Publish internal research notes on experiments that did not ship — failure docs matter here
- Collaborate with MLOps to ensure every model has a monitoring and drift-detection strategy
[ WHAT WE'RE LOOKING FOR ]
- MSc or PhD in Statistics, Machine Learning, Computer Science, or a quantitative field
- 3+ years professional ML engineering with models running live in production
- Deep familiarity with PyTorch or JAX; scikit-learn for classical models
- Experience with Bayesian methods (PyMC, Stan, or equivalent)
- Python proficiency: type hints, testing, clean module structure — no notebooks in production
- Strong grasp of experiment design and statistical rigour
[ NICE TO HAVE ]
- Published work in forecasting, causal inference, or multi-agent RL
- Experience with marketing mix models or media optimisation
- Background with MLflow, Weights & Biases, or equivalent experiment tracking
[ APPLY ]
Apply for this role