Job Role:
A dedicated startup is being formed to industrialize and scale a secure, AI-enabled, multi-source decision-support software offering. The platform is a multi-sensor fusion and agentic AI solution connecting to diverse data sources (for example geospatial layers, imagery, video, and other operational signals). This role will support the delivery of a scalable product and contribute to establishing the processes, standards, and collaboration practices required for sustainable growth.
Own the reliability and scalability of ML and LLM-enabled services by building robust pipelines, deployments, monitoring, and operational controls in a fast-moving startup environment.
Job Responsibilities:
- Design and operate end-to-end ML/LLM delivery pipelines: data to training/fine-tuning to evaluation to packaging to deployment
- Build CI/CD for models and services, including automated testing, validation gates, and rollback strategies
- Standardize experiment tracking, model/version lineage, and artifact management (datasets, prompts, checkpoints, embeddings)
- Implement monitoring and observability: latency, cost, drift, quality signals, and safety/guardrails metrics
- Optimize inference performance and cost (batching, caching, quantization, hardware choices)
- Define and enforce environment and dependency management across dev/stage/prod
- Work with engineering on scalable serving patterns (APIs, streaming, event-driven), and with security on access controls and secrets
- Support release readiness: runbooks, incident response, SLOs/SLAs, and post-release stability tracking
- Coordinate with procurement and legal where needed for tooling, cloud services, and vendor onboarding
- Startup mode: hands-on, flexible, comfortable pivoting, and able to unblock teams quickly
- Interfaces / stakeholders
- Software engineering (platform, backend, DevOps)
- ML/LLM engineers and applied scientists
- Product and delivery teams (PM/PO/BA)
- Security, IT, procurement, and finance (as applicable)
Qualifications and Experience:
- Typically 5+ years in MLOps/DevOps/Data Platform roles, including production deployments of ML and/or LLM-powered systems. Experience in fast-paced product environments preferred.
- Tools (examples)
- ML lifecycle: MLflow / Weights & Biases / equivalent
- Serving: FastAPI, Triton (plus), Ray Serve (plus)
- Orchestration: Airflow/Dagster (plus)
- Observability: Prometheus/Grafana, OpenTelemetry, ELK
- Cloud: AWS/Azure/GCP (or private cloud)
- KPIs
- Deployment frequency and lead time for model releases
- Production stability: incident rate, MTTR, SLO compliance
- Model quality health: drift detection coverage, evaluation gate pass rate
- Inference cost and latency improvements
- Reproducibility and traceability coverage (lineage completeness)
