Senior ML Engineer

Contenido del rol extraído en secciones para revisar más rápido.

Own the client's recommendation system end-to-end: design, training, deployment, and monitoring
Build and mature ranking and retrieval models that process behavioral signals — clicks, opens, purchases — at consumer scale
Operate ML systems at hundreds of millions of events per day, with direct impact on client revenue
Make independent technical decisions and engage directly with client engineering leadership — no intermediaries

6 to 9 years of experience in software or data engineering roles
4+ years building and operating ML systems in production
2 to 3+ years with recommendation systems as a core area of work — not adjacent exposure
Hands-on experience with ranking and retrieval architectures; able to discuss tradeoffs from real production work
Direct experience operating systems at consumer scale (hundreds of millions of events per day or more)
Strong Python skills and fluency with a modern ML stack: PyTorch or TensorFlow, scikit-learn, feature engineering libraries
SQL fluency for data exploration and validation
Production deployment experience on AWS, GCP, or Azure
Fluent spoken and written English — this role involves direct technical conversation with US client engineering leaders
Demonstrated autonomy in client or stakeholder-facing settings: owning scope, raising blockers, making independent decisions

Background in marketing technology, e-commerce, ad tech, or messaging platforms
Experience with multi-tenant or multi-brand data structures
Familiarity with feature stores (Feast, Tecton, or in-house) and low-latency online inference
Experience designing A/B tests and communicating results in business terms
Hands-on experience with Spark or distributed data processing frameworks

Your ML work has been primarily in NLP, computer vision, or fraud detection, with recsys as a secondary exposure
The largest production system you've operated handled thousands or low millions of events per day
You're a research-oriented profile without ownership of production systems
Your English isn't ready for unscripted technical conversation