This is a fully remote position, open to applicants in United States, +4 more locations.

📋 Description

• Develop and manage the real-time inference service responsible for scoring models in the risk decision engine, focusing on low latency and high availability.

• Take charge of the model deployment infrastructure, encompassing registry and versioning, model CI/CD checks, shadow mode, and staged rollouts.

• Create model observability metrics for availability, latency, errors, and drift detection to trigger retraining.

• Collaborate with Risk Data Science to transition models from development to production operation under MLP ownership.

• Establish experimentation functionalities, including champion/challenger and canary routing.

• Implement explainability outputs such as SHAP attributions.

• Assume product ownership, self-organize on projects, and contribute to shaping a new platform team.

⛳️ Requirements

• Over 5 years of experience in machine learning engineering, backend software engineering, MLOps, or a closely related discipline.

• Experience with production ML services: deploying, serving, and managing models within low-latency, high-availability environments.

• Solid backend engineering skills in Python, particularly with API frameworks like FastAPI or Flask.

• Proficient in model deployment and lifecycle management tools: model registries, CI/CD for models, versioning, and staged rollout strategies (shadow, canary, champion/challenger).

• Experience in building observability and alerting systems for production services, focusing on latency, errors, and ideally model-specific signals such as drift.

• Familiarity with SQL, key-value/low-latency stores like Redis or DynamoDB, and streaming technologies such as Kafka, Kinesis, or Redpanda.

• Knowledge of a modern data stack, including Snowflake, dbt, Dagster, or Airflow.

• Experience working in a regulated, audit-sensitive, or compliance-oriented environment.

• Exposure to functional programming languages or a willingness to work with Haskell, React, and TypeScript.