Senior Machine Learning Engineer

Job Title- PyTorch & Python ML Engineer

Location- Mumbai/Bangalore/Hyderabad/Gurgaon/Indore

Mode- Onsite

Job Type: Full-Time

Core Responsibilities

The role designs and owns end-to-end ML pipelines implemented in Python, covering audio data ingestion, preprocessing, feature extraction, training invocation, evaluation, and artifact generation.

The role is responsible for implementing and maintaining large-scale speech models using PyTorch, including transformer-based architectures for speech-to-text and text-to-speech, ensuring correctness, performance, and extensibility.

The role owns audio data ETL and dataset management workflows, including data validation, labeling interfaces, dataset versioning, and traceability between datasets and trained models.

The role ensures that model implementations are training- and inference-compatible, avoiding mismatches in tensor shapes, precision, preprocessing, or runtime assumptions.

The role works closely with training orchestration teams to ensure that model code is structured for distributed training, checkpointing, and resumability.

The role collaborates with inference and GPU teams to export trained models into production-ready formats and integrate them into high-performance inference services.

The role establishes coding standards, testing practices, and performance guidelines for Python-based ML code to ensure maintainability over long platform lifecycles.

Operational Ownership

The PyTorch & Python ML Engineering Lead owns ML implementation failures. If a model underperforms or behaves incorrectly due to bugs, mismatched preprocessing, or incorrect integration, this role is accountable.

The role owns the model lifecycle integrity, ensuring that trained artifacts can always be traced back to data, code versions, and training configurations.

The role is accountable for ensuring that ML pipelines are repeatable and debuggable, especially when issues arise in production.

The role participates in production incident analysis when issues stem from model implementation, data handling, or ML pipeline Behaviour.

Key Interfaces

This role works closely with the Distributed Systems Architect and Technical Governance & Platform Solutions Lead to ensure ML workflows align with architectural and design constraints.

The role interfaces daily with the PyTorch Lightning Training Lead to ensure that model code is structured correctly for scalable training workflows.

The role collaborates with the Speech Modelling & Quality Lead to translate quality improvement goals into concrete model changes and data requirements.

The role works with the GPU Inference Optimization Lead to ensure model artifacts are compatible with inference runtimes, precision requirements, and latency budgets.

The role coordinates with GitOps, CI/CD & Reliability teams to ensure ML pipelines integrate cleanly with build, deployment, and observability systems.

Explicit Non-Responsibilities

This role does not define speech quality targets or language-level accuracy goals; those are owned by the Speech Modelling & Quality Lead.

This role does not manage GPU infrastructure, Kubernetes scheduling, or inference service scaling.

This role does not own product-level feature definition or API contracts.

Role Expectation

The PyTorch & Python ML Engineering Lead is expected to ensure that machine learning development is treated as production engineering, not experimentation.

Success in this role is measured by:

Stable, maintainable ML codebases

Reproducible training and evaluation outcomes

Smooth transition from training to inference

Minimal production issues caused by ML pipeline defects

Ability to support rapid model iteration without system instability