AI/ML engineer
You'll take machine learning models from a data scientist's notebook to production - MLOps, model serving, and the evaluation discipline that keeps them honest.
About the role.
You'll work alongside our AI & Data practice to take models from a data scientist's notebook into production systems that enterprise clients actually run - which means MLOps, model serving, monitoring, and the evaluation discipline that catches drift before a client does.
Clients bring genuinely hard problems: fraud detection at transaction-time latency, demand forecasting across volatile supply chains, document extraction pipelines that have to be right, not just plausible.
Apply for this roleBengaluru / Pune / Toronto
The shape of the work.
Productionise models
Turn research-quality models into services with defined SLAs - versioned, monitored, and rollback-able like any other production system.
Own the evaluation harness
Build and maintain the offline and online evaluation suites that tell us - before the client does - when a model's performance has degraded.
Work directly with client data science teams
Pair with client-side practitioners to transfer the operational discipline, not just hand over a deployed endpoint.
What we're looking for.
2+ years shipping ML in production
Experience taking at least one model beyond a notebook - training pipelines, serving infrastructure, monitoring.
Strong Python
Comfortable across the modern ML stack - PyTorch or TensorFlow, plus the surrounding tooling (MLflow, Airflow, or equivalents).
MLOps platform experience
Kubeflow, SageMaker, Vertex AI, or a comparable production ML platform.
Domain exposure
Prior work in fraud, forecasting, or document/NLP pipelines is a strong plus.