Machine Learning Engineer Sr

Role Summary

Leega is a company focused on efficient and innovative service to its customers.

This couldn't be any different with our main fuel: people!

Our culture is inspiring and our values are present in everyday life: ethics and transparency, quality excellence, teamwork, economic, social and environmental responsibility, human relations and credibility.

We are looking for innovative professionals who are driven by challenges and focused on results.

If you are looking for a dynamic and partner company that invests in its employees through constant training, Leega is the place for you!

>> LEEGA IS FOR EVERYONE, we will be very happy to have you on our team. Come and be part of our history and the construction of our future.

Sign up for our vacancies right now!

You are the one who connects the prototype to production. Will design and build the ML engineering of the pricing engine — the inference serving, the training pipelines and the feature engineering — so that complex models run in real time, with low latency, on Ray. Focuses on modeling and ML code; the platform and runtime are with the MLOps/Platform team, with whom you work side by side.

Your Challenges

Inference serving — designing the chained model pipeline on top of Ray Serve — model composition, low latency, and update strategies.

Distributed training — build training pipelines (Ray Train/Data), HPO (Ray Tune) and tenant-trained models, with resilient checkpointing.

Feature engineering — define and materialize features in the feature store (Feast/Redis), ensuring consistency between training and production.

Optimization and RL — implement and optimize the optimization (linear programming) and offline RL components of the pricing pipeline.

Model quality — monitor drift from a modeling perspective, validate versions, and produce explainability (SHAP) — in partnership with MLOps.

Technical leadership — act as a reference, mentor and define, with the team, what is viable and scalable.

You handoff with data scientists, receive data from data engineers and hand it over to the MLOps/Platform team for deployment and operation.

• Stack & Tools
• ML serving & training: Ray (Serve, Train, Tune, Data, RLlib)
• Registry & features: MLflow, Feast + Redis
• Optimization: linear programming (Gurobi, HiGHS), offline RL
• Language & runtime: Python; Docker; Iceberg reading
• What We Look For
• Essentials

Proven experience putting ML models into production.

Python and solid Software Engineering fundamentals (APIs, tests, clean code).

Serving and inference optimization for low latency.

Familiarity with containers (Docker) and MLOps flows (registry, deployment).

Comfort with AI-assisted development (Claude Code).

Differences

Ray ecosystem (Serve, Train, Tune, RLlib) — strong differentiator.

Feature stores (Feast) and low latency serving with Redis at scale.

Optimization/solvers (Gurobi, HiGHS) or revenue management in real time; RL offline.

Serving generative AI (vLLM, LiteLLM) and multi-tenant architectures.

Benefits

Remote Work

Project Time: 6 months, with the possibility of extension/internalization.