Lead Software Platform Engineer, MLOps

Role Overview

Lead Software Platform Engineer (MLOps) at the intersection of distributed systems and MLOps. Own and scale the AI/ML platform that customers use to run production-grade AI/ML workflows, and that other engineering teams build against.

What You Will Do

• Own the technical architecture of the AI/ML platform: service and API surface for running models and agents on customers’ scientific data.
• Own the end-to-end model and prompt lifecycle across Databricks MLflow and AWS Bedrock (registration, versioning, asset bundles, staged promotion, rollback, and multi-model serving).
• Design inference substrate for real-time and batch workloads: routing, batching, caching, concurrency control, GPU/accelerator capacity planning, handling large binary inputs (e.g., instrument images), and graceful degradation under load.
• Integrate models and LLMs into production systems using RAG, and architect the agentic layer (tool/function calling, MCP-based tooling, and agent runtimes), deciding what belongs in the platform vs. applications built on top.
• Build security into the platform: guardrails, prompt-injection/tool-abuse defenses, PII/PHI handling, and hard tenant data boundaries across prompts, retrieval, and tool calls (with the security team).
• Build evaluation and quality infrastructure to make AI shippable (offline and online evaluation and observability).