Lead Data Scientist – Sourcing Research

About the role

Join Coupa’s Sourcing research team in Prague to lead end‑to‑end data science initiatives that shape how millions of buyers and suppliers interact. You will work on greenfield problems, building models from scratch on a unique dataset of $10 trillion in spend and 10 M+ entities.

Key responsibilities

• Own sourcing research projects from problem framing through production deployment.
• Design datasets, labels and benchmarks that turn historic spend events into learnable training data.
• Develop and experiment with a variety of methods – neural networks, gradient‑boosting, optimisation, bandits, mechanism design – selecting the best tool for each problem.
• Leverage Rossum’s proprietary T‑LLM architectures to extract signal from multimodal documents such as RFPs, quotes and contracts.
• Ship models with production constraints in mind, ensuring low latency, robustness and safe failure handling.
• Collaborate across Coupa’s Sourcing product, data teams and AI Platform to integrate research into live services.

Required profile

• 5+ years of applied ML, data science, ML engineering or quantitative research experience, with productionised models.
• Strong Python skills and proven ability to work with large, messy datasets, including SQL.
• Deep expertise in at least one modelling discipline (e.g., deep learning, recommendation, forecasting, optimisation, causal inference, RL, LLM‑based systems).
• Scientific rigor in experiment design, metric evaluation and leakage prevention.
• Ownership mindset and curiosity to thrive in ambiguous, greenfield problem spaces.

Required skills

• Python
• SQL
• Machine learning
• Deep learning
• Recommendation and ranking systems
• Forecasting and time‑series analysis
• Optimization / operations research
• Causal inference and econometrics
• Reinforcement learning and bandits
• Mechanism design
• Large language models (LLM)

What we offer

• Greenfield problems on a mature product with $10 trillion of spend data.
• Proprietary in‑house T‑LLM models you will help evolve.
• Direct impact: see your research deployed to real buyers.
• State‑of‑the‑art compute resources, including high‑end GPUs and large‑memory clusters.
• Budget for conferences and professional development.
• 33 days of paid time off plus parental leave and wellness days.
• In‑office workspace in Prague’s Karlín district with a terrace view of Prague Castle.