ML Engineer AI COE

ML Engineer AI COE Hybrid Model Location - Hyderabad Apply ML/AI solutions with awareness of business needs, system constraints, and business context Build and own ML/DL models across complex data types — geometries, part metadata, transactional data, and free-text notes Contribute NLP and document understanding pipelines for technical drawings and unstructured manufacturing specs; build reusable components on our AWS Bedrock-based AI Platform Tackle different complex ML problems: identify data issues, navigate model choices, and design clear experiments Own small–medium ML/DL subsystems and features end-to-end Work independently on assigned ML tasks while collaborating across teams Suggest improvements at the feature level; explore and evaluate new techniques with guidance Mentor junior peers to grow into ML; provide solid code, testing, and reviews Contribute to feature-level design discussions and surface technical suggestions and improvements Must-Have Requirements Good grounding in the mathematical foundations of ML and a relevant degree (computer science, simulation science, or equivalent) 3–5 years of hands-on experience designing, training, and deploying complex ML/DL models in production using state-of-the-art frameworks (PyTorch, TensorFlow, scikit-learn, XGBoost) Proven ability to build models that serve as autonomous decision systems— applied to critical business problems with large, heterogeneous data Strong Python engineering fundamentals: clean, modular, testable code; data pipelines; model versioning; CI/CD for ML; packaging for production (e.g. Docker, service wrappers) Solid ML experimentation practice: experiment tracking from scratch (e.g. W&B, MLflow), model evaluation, and iterative improvement tied to measurable business outcomes Solid understanding of data transformation techniques, languages, and libraries (e.g.

Pandas, Polars, SQL, dbt) Ability to work independently: break down larger problem definitions into concrete tasks, navigate model choices, and deliver production-ready ML implementations end-to-end