ROLE OVERVIEW Lead the engineering platform and operating model for transformer-based foundation models trained on large-scale customer-event data. This role owns the architecture for cloud infrastructure, orchestration, release controls, observability and scalable execution across pretraining, fine-tuning and scoring. It combines hands-on system design with project management, cross-functional leadership and accountability for secure, reliable and cost-conscious delivery.

KEY RESPONSIBILITIES
• Define the target architecture and engineering roadmap for transformer training, evaluation, registry, deployment and scoring workflows across cloud environments.
• Lead orchestration design for long-running, dependency-heavy and restartable pipelines, including distributed GPU workloads and controlled promotion between environments.
• Establish infrastructure patterns for compute, storage, networking, IAM, secrets, encryption, workplace isolation, backup and disaster recovery.
• Own CI/CD, Infrastructure as Code, testing strategy, artefact management, model lineage, approval gates and rollback mechanisms.
• Design observability for pipeline health, model and data drift, resource utilisation, latency, failures and cloud spend; define measurable service objectives.
• Partner with data scientists to optimise throughput and reliability without compromising validated model behaviour, reproducibility or governance evidence.
• Lead engineers through system-design reviews, coding standards, incident analysis, documentation, mentoring and technical decision-making.
• Create delivery plans, estimate capacity, manage risks and dependencies, and coordinate cloud, cyber security, data, validation and business stakeholders.
• Provide concise updates to senior stakeholders and drive production-readiness reviews, release decisions and post-deployment improvement plans. REQUIRED SKILLS AND

EXPERIENCE:
Programming: Strong Python and SQL skills,



with production-quality coding, testing, packaging and Git-based development practices Pipeline orchestration: Hands-on experience designing and operating training, evaluation, batch-scoring and deployment pipelines using tools such as Airflow, Kubeflow, Azure ML, AWS SageMaker or Vertex AI Cloud infrastructure: Practical knowledge of cloud compute, storage, networking, IAM, secrets management and GPU-enabled workloads on Azure, AWS or GCP MLOps: Experience with Docker, CI/CD, artefact and model registries, MLflow or Weights & Biases, configuration management and reproducible environments Transformer operations: Understanding of distributed or multi-GPU training, checkpointing, mixed precision, data loading, model serving and performance optimisation for transformer-based models Reliability: Experience with monitoring, logging, alerting, lineage, rollback, cost controls, security and production-support practices Architecture and leadership: Proven system-design depth for ML platforms and experience leading engineering workstreams, technical reviews, delivery planning and stakeholder engagement GOOD TO HAVE:
• Infrastructure as Code using Terraform, Bicep or CloudFormation and Kubernetes-based platform engineering.
• Distributed training and compute using PyTorch Distributed, DeepSpeed, Ray, Spark or managed cloud equivalents.
• Model-serving optimisation, autoscaling, feature stores, event streaming and low-latency inference patterns.
• Security and governance for regulated financial-services workloads, including auditability and segregation of duties.
• Cloud architecture, DevOps, Kubernetes or machine-learning certifications.

QUALIFICATIONS:
• Bachelor’s or Master’s degree in Computer Science, Engineering, Data Science or another relevant technical discipline.
• 4–8 years of ML engineering, cloud platform, DevOps or data-platform experience, including production ML systems. Demonstrated ownership of cloud and MLOps architecture plus leadership of complex cross-functional deliveries.

📌 Senior Machine Learning Engineer (Gurugram)
🏢 EXL
📍 Gurugram