thyssenkrupp Materials Services - ML/Agent Ops Engineer

This role operates production AI, ML and agentic workloads, with a focus on reliable deployment, observability, evaluation and controlled releases. The posting lists Mumbai and Thane as locations.

Responsibilities

• Design CI/CD pipelines for models, prompts, agents and supporting infrastructure across development, test and production. Automate deployments, versioning, rollbacks and environment promotion, and implement runtime safeguards.
• Establish tracing, monitoring and alerting for AI applications and agents, including token-consumption analysis, latency tracking and incident diagnostics.
• Build evaluation pipelines and acceptance gates covering quality, groundedness, task adherence, safety and agent behavior. Manage prompt lifecycles, optimize RAG and semantic retrieval, and integrate vector- or search-based knowledge components where needed.
• Work with security, engineering and data teams on identity, secrets management, compliance controls and cost optimization, balancing reliability, security and performance.

Requirements

• At least 5 years of experience in DevOps, platform engineering, MLOps or a closely related role. The structured posting lists a bachelor's degree.
• Experience operating production cloud workloads using CI/CD, monitoring and infrastructure automation; strong experience with Azure, GitHub or Azure DevOps, Docker, Kubernetes, Terraform or Bicep, and infrastructure-as-code practices.
• Hands-on MLOps, LLMOps or AgentOps experience covering deployment, monitoring, retraining or reevaluation, and controlled releases. Strong understanding of logs, metrics, traces, runtime telemetry and production AI diagnostics; practical Python skills for automation, tooling and evaluation orchestration. Familiarity with retrieval-augmented systems, prompt engineering, tool-calling flows and agent debugging.

Preferred

Experience with production AI, ML or agentic workloads is strongly preferred; experience in high-availability, regulated or enterprise-scale environments is an advantage.