ML Platform Engineer

Build the data infrastructure that powers robots in the real world. Robotics is moving from research labs into production fleets across factories, warehouses, vehicles, defense systems, agriculture, logistics, and field deployments. As robots scale across the physical world, every failure, regression, edge case, and unexpected behavior becomes a data problem: what happened, when, on which robot, and why? Every robot, in every industry, requires the same core capabilities: to sense, understand, and act on multimodal data from the physical world. At Foxglove, we built the agentic data platform robotics and Physical AI teams use to answer those questions. We help robotics teams make vast quantities of robot data actionable, creating the data flywheel they need to develop, test, train, deploy, and operate robots with confidence. About the Role We're looking for a ML Platform Engineer with deep infrastructure instincts to help design, deploy, and scale the systems that power Foxglove's data platform. This is a platform-first role: you'll own the infrastructure layer that makes ML possible in production, not just the models that run on top of it. You'll be responsible for the reliability, scalability, and performance of the ML platform itself, from inference serving and pipeline orchestration to training infrastructure and evaluation frameworks. The problems are real and urgent: petabyte-scale multimodal robotics data, high-throughput retrieval and embedding pipelines, and the internal ML flywheel that lets our team ship fast. This is a hands-on infrastructure role, not research. Key Responsibilities - Design, deploy, and operate production inference infrastructure — including model serving, autoscaling, load balancing, and cost optimization across cloud environments - Own the platform architecture for embedding and retrieval pipelines that power semantic search over multimodal robotics data (image, video, point cloud, and timeseries) - Build and maintain the training and evaluation infrastructure that enables rapid iteration on model performance — including job orchestration, experiment tracking, and dataset versioning - Drive cloud infrastructure decisions (AWS/GCP) that directly impact latency, throughput, reliability, and cost at scale - Define platform abstractions and internal tooling that let product engineers ship ML-powered features without needing to manage infrastructure themselves - Evaluate, integrate, and operationalize third-party ML infrastructure components; establish clear build vs. buy frameworks for the team What We're Looking For - Deep, hands-on experience owning production ML infrastructure: inference serving, model optimization (e.g., vLLM, Triton, TorchServe), orchestration, and cloud cost management - Strong foundation in distributed systems and cloud infrastructure (AWS/GCP) — you think in terms of system reliability, failure modes, and operational burden, not just model accuracy - Experience architecting and operating retrieval systems at scale, including vector databases (e.g., Pinecone, Lance, turbopuffer, pgvector) and embedding pipelines over large, heterogeneous datasets - A platform engineer's mindset: you build systems that other engineers depend on, and you take that responsibility seriously - Proven ability to operate with high ownership — you can make hard infrastructure tradeoffs independently and move fast without breaking things - Strong communication skills; you can explain infrastructure tradeoffs clearly to both ML and non-ML engineers Bonus Points - Familiarity with fine-tuning and domain adaptation techniques for LLMs or embedding models (i.e. SFT, PEFT) - Familiarity with data mining or hybrid search workflows, especially as applied in robotics autonomous vehicles, or physical AI workflows - Prior experience building ML platforms, evaluation frameworks, or data management tooling from the ground up What We Offer - $300 monthly budget towards commuter benefits or building your personal w