Applied ML Engineer

Physical AI is moving from research labs into production fleets across industries. As robots scale across the real world, from factories to vehicles, to defense - every workflow from product development to deployment becomes a data problem: what happened, when, on which robot, and why?

At Foxglove, we built the unified data platform for physical AI that developer and engineering teams use to answer those questions. We help teams make vast quantities of robotics data actionable, creating the data flywheel they need to develop, test, train, deploy, and operate robots with confidence.

We're looking for an Applied ML engineer with deep infrastructure instincts to help design, deploy, and scale the ML systems that power Foxglove's data platform.

In this role, you'll own the infrastructure that makes ML work in production: from optimizing inference pipeline throughput to standing up training and eval workflows. You'll work directly on the problems that matter right now: retrieval applications over petabyte-scale multimodal robotics data, using the latest models to build high-performance search and data mining products, and creating the internal ML flywheel that lets us iterate fast. This is a hands-on application-driven role, not research.

• Deploy and operate inference infrastructure for production ML workloads, including model serving, scaling, and cost optimization

• Build and maintain vector database integrations and embedding applications to support semantic search over multimodal (image, video, point cloud, and timeseries) robotics data

• Design and implement evaluation and training infrastructure, to help us iterate quickly on model performance

• Own cloud architecture decisions and tooling that affect inference latency, throughput, cost, and reliability at scale

• Collaborate with product engineers to ship application-driven ML features tailored to developers building the cutting edge of robotics and physical AI, not prototype experiments

• Identify the right off-the-shelf solutions and adapt them for production, and know when to build vs. buy

• Strong hands-on experience in production ML infrastructure: cloud inference, model serving optimization frameworks (e.g., TorchServe, vLLM, Triton), and cost management

• Experience with the technologies used in building retrieval systems, including vector databases (e.g., Pinecone, Lance, turbopuffer, pgvector) and text-image embedding models

• Solid engineering fundamentals: distributed systems, cloud infrastructure (AWS/GCP), and production reliability

• A bias toward application and product impact over research; you’re excited by shipping things that work, not writing papers

• Proven ability to operate independently, make good tradeoffs, and move fast in a high-ownership environment

• Excellent communication skills; you can explain ML tradeoffs to non-ML engineers

• Familiarity with fine-tuning and domain adaptation techniques for LLMs or embedding models (i.e. SFT, PEFT)

• Experience with data mining or hybrid search workflows, especially as applied in robotics autonomous vehicles, or physical AI workflows

• Experience building ML tooling, data management, and evaluation frameworks from scratch

• Work on real robotics problems. Robot data is large, messy, multimodal, time-sensitive, and tied to physical-world behavior. The problems we work on span ingestion, indexing, search, visualization, replay, connectivity, collaboration, evaluation, and operations.

• Build tools engineers rely on. Foxglove is used by robotics teams investigating failures, validating changes, reviewing field behavior, curating datasets, and operating production fleets. The work you do helps teams understand what their robots saw, what they did, and why they behaved the way they did.

• High-leverage product surface area. A better query path, visualization workflow, Fleet connection, UI primitive, API, onboarding flow, or customer deployment can change how an entire robotics team works.

• Ownership and autonomy. We’re a small team, and people at Foxglove own meaningful work end-to-end. You’ll have real influence over product direction, technical architecture, customer outcomes, and how we operate as a company.

• Strong peers and high standards. You’ll work with people who care about correctness, performance, craft, product judgment, and building software that technical users trust under pressure.

• A mission grounded in production software. We accelerate robotics and physical AI by building the infrastructure teams use every day to connect to robots, inspect live telemetry, manage multimodal data, replay runs, investigate failures, and improve real systems.

• Competitive equity grant in a Series B company.

• Medical, dental, vision, and term life insurance coverage at 100% for employees and 75% for dependents, for U.S. full-time employees.

• 401(k) matching up to 4%, for U.S. full-time employees.

• 4 weeks of vacation, plus holidays and winter break.

• All-expenses-paid company offsites 1–2× per year.

• $300 monthly budget toward commuter benefits or building your personal workspace, depending on role/location.

Foxglove is an equal opportunity employer. We welcome candidates from different backgrounds, experiences, and communities, and we’re committed to building an inclusive environment for everyone.

We encourage you to apply even if you don’t meet every nice-to-have listed above. The strongest candidates often bring a mix of relevant experience, curiosity, judgment, and the ability to learn quickly.

Foxglove is the data platform for Physical AI. Built for robotics teams developing real-world systems, Foxglove provides a purpose-built, modular platform to collect, organize, and learn from vast quantities of multimodal data, creating the data flywheel to safely scale from development to distributed fleets. Founded in 2021, Foxglove supports hundreds of customers across automotive, aerospace, defense, logistics, agriculture, construction, and consumer robotics to deploy the next generation of intelligent machines. Learn more at foxglove.dev.