Overview

Myntra is India’s leading fashion and lifestyle platform, where technology meets creativity. As pioneers in fashion e-commerce, we’ve always believed in disrupting the ordinary.

Responsibilities

• Build developer-facing platforms, SDKs, APIs and abstractions.
• Reduce the time required to move an ML experiment into production.
• Eliminate repetitive infrastructure work for Data Scientists and ML Engineers.
• Build standardized templates and paved roads for ML development.
• Improve debugging, discoverability and observability of ML workloads.
• Enable teams to focus on models and business problems rather than infrastructure.
• Own architecture and technical direction for major ML Platform initiatives.
• Lead complex system-design discussions and technical reviews.
• Convert ambiguous problems into scalable platform solutions.
• Drive engineering excellence across reliability, scalability, performance and maintainability.
• Mentor engineers and raise the technical bar of the team.
• Influence architecture across ML, Data, Platform, SRE and Infrastructure teams.
• Evaluate emerging technologies and make pragmatic build-vs-buy decisions.
• Take critical systems from concept through architecture, implementation and production adoption.

Qualifications

• Architect, build and evolve scalable ML Platform and MLOps capabilities for large-scale production workloads.
• Create self-service infrastructure that allows Data Scientists and ML Engineers to train, experiment, deploy and operate models without managing underlying infrastructure.
• Design reusable platform abstractions, SDKs, APIs and tooling that dramatically improve ML developer productivity.
• Build systems for reproducibility, lineage, versioning, governance and lifecycle management across ML workflows.
• Drive architecture and technology choices for Myntra's ML infrastructure.
• Build infrastructure for large-scale distributed training across CPU and GPU clusters.
• Design and operate multi-node and multi-GPU training environments.
• Work with distributed computing frameworks such as Ray or equivalent technologies.
• Build intelligent scheduling, resource isolation and autoscaling capabilities for ML workloads.
• Improve utilization of expensive GPU infrastructure through scheduling, workload optimization and capacity management.
• Design fault-tolerant training systems with checkpointing, retries and recovery mechanisms.
• Build a world-class training platform supporting the complete ML development lifecycle.
• Design scalable orchestration for training, feature engineering, validation and deployment workflows.
• Build reusable workflow components using technologies such as Airflow, Ray and Kubernetes.
• Improve scheduling, dependency management, execution isolation and reliability for thousands of ML workloads.
• Enable experimentation across different compute environments without exposing infrastructure complexity to users.
• Build end-to-end ML lifecycle capabilities covering:
• Build experiment tracking and model management using MLflow or equivalent technologies.
• Enable reliable model versioning, approval, rollout and rollback.
• Build automated model validation and production-readiness workflows.
• Enable reproducible ML workflows across development, staging and production environments.
• Build highly scalable infrastructure for real-time, batch and asynchronous model inference.
• Design model-serving platforms running on Kubernetes and GPU infrastructure.
• Optimize serving systems for latency, throughput, availability and cost.
• Explore and adopt technologies such as Ray Serve, NVIDIA Triton, vLLM, SGLang or equivalent platforms where appropriate.
• Enable production deployment of traditional ML, deep-learning and emerging AI/LLM workloads.
• Treat ML infrastructure as a production-grade distributed platform.
• Define and drive SLIs, SLOs, availability and reliability standards for ML platform services.
• Build deep observability across infrastructure, pipelines, training workloads and inference systems.
• Troubleshoot challenging production issues spanning Kubernetes, GPU workloads, distributed systems, networking, storage and ML pipelines.