Requirements

Strong software engineering background building reliable, scalable production systems (Python preferred)
Hands‑on experience supporting large‑scale ML / LLM training, evaluation, or experimentation infrastructure
Operating GPU‑heavy workloads in cloud environments using Docker and Kubernetes (scheduling, utilization, isolation)
Designing and running data / compute pipelines and orchestration (e.g., Airflow, Argo) with object storage (Azure Blob / S3)
Platform reliability and operability: observability, metrics, logging, tracing, alerting (Prometheus, Grafana, OpenTelemetry)

Nice to Haves

Building secure, reproducible platforms using CI/CD, infrastructure‑as‑code (Terraform, Helm), container security, and secrets management
Experience working closely with AI researchers in fast‑moving, experimental, frontier‑scale research environments and building internal tools (CLIs, portals, APIs) to boost productivity

Design and build core platform services for scalable training and evaluation, including cluster orchestration, job scheduling, data and compute pipelines, and artifact management.
Standardize containerized workflows by maintaining Docker images, CI/CD, and runtime configurations; advocate for best practices in security, reproducibility, and cost efficiency.
Implement end-to-end observability and operations through metrics, tracing, logging, dashboard development, monitoring, and automated alerts for model training and platform health (using Prometheus, Grafana, OpenTelemetry).
Architect and operate services on Azure cloud platforms, managing infrastructure-as-code (Terraform/Helm), secrets, networking, and storage.
Enhance developer experience by creating tools, CLIs, and portals that simplify job submission, metrics analysis, and experiment management for generalist software engineering and research teams.
Enforce security and compliance policies for data access, container hardening, and supply-chain integrity, and partner with security and privacy teams to maintain robust practices in multi-tenant environments and secret management.
Collaborate cross-functionally with data, model, and product teams to align infrastructure roadmaps with training needs, evaluation protocols, and Copilot product goals.