Relational Foundation Model Engineer, Modern Data Stack

As an engineer on this team, you won’t just be fine‑tuning existing models; you’ll be designing and experimenting with novel Transformer and GNN architectures that generalize across diverse relational schemas. Your work will directly impact real‑world applications spanning recommendation systems, demand forecasting, fraud detection, and predictive maintenance — all powered by one extensible model. You’ll collaborate closely with researchers and engineers across the full ML lifecycle, from architecture exploration and large‑scale training to post‑training optimization and inference acceleration.

This is a rare opportunity to contribute to foundational research that ships into production and shapes the modern data stack. If you’re excited about graph learning, relational reasoning, and building AI systems that go far beyond single‑table benchmarks, this is the team for you.

What you’ll be doing:

• Collaborate with researchers/engineers to enhance our Transformer and GNN‑based models to operate seamlessly over any relational schema and heterogeneous graph.
• Gain hands‑on experience with high‑impact use cases such as forecasting, entity matching, customer retention and fraud detection – all built on top of a single, extensible foundation model.
• Leverage your knowledge in ML and AI to tackle real challenges while contributing to scalable and adaptable solutions that push the boundaries of what’s possible.
• Work may span the full lifecycle of modern ML systems: from architecture design/training to post‑training optimization and inference acceleration.
• Contribute to our next generation of the Relational Foundation Model.

What we need to see:

• MS or PhD in Machine Learning, Computer Science, or equivalent program.
• Proficiency in Python and deep learning frameworks, such as PyTorch.
• At least 8 years of research experience in designing ML algorithm solutions.
• Practical experience in using Predictive Models in Real World Applications.

Ways to stand out from the crowd:

• Familiarity with graph‑based machine learning.
• Publications at venues such as NeurIPS, ICLR, ICML, or similar.