Head of Applied ML

Anvil Robotics•Taipei, CA

About The Position

Anvil Robotics is building the Physical AI platform for robotics builders—modular hardware, software, and data tooling. We're hiring a Head of Applied ML to eliminate the biggest bottleneck our customers face: getting custom models deployed on real hardware without requiring Anvil engineering support. This is not a research role. You're not pushing the frontier—you're making the frontier accessible. Your job is to stay on top of published research and off-the-shelf models, identify what's most high-leverage for our customers, and get it working on Anvil hardware so customers don't have to figure it out themselves. Every initiative you take on needs to produce something tangible and useful within 100 days—no exceptions. This role is also deeply communicative in two directions. Internally, you're the person who keeps Anvil's leadership informed about what's happening in robot learning and what it means for our product and customers—in plain language, not jargon. Externally, you're working directly with deployment customers to understand their constraints and help them get to working outcomes. Both audiences are non-technical relative to you, and making them smarter is part of the job.

Requirements

Hands-on experience deploying Generative AI robotics models to real hardware
Dogs, cobots, humanoids, Diffusion Policy
VLAs, Behavior Cloning, Robot Foundation Models, World Models, Physical Robot RL
Strong familiarity with the robot learning research landscape—you follow the work coming out of leading labs and can quickly assess practical applicability
Experience working directly with customers or end users to understand deployment needs
Ability to implement and adapt published research or off-the-shelf models, not just run existing pipelines

Nice To Haves

You love keeping up with robot learning research and ML developments, but you're wired toward application—you're always asking "how can I use this for someone right now?"
You think in delivery windows, not research arcs—nothing you take on goes past 100 days without a tangible output in hand
You're an exceptional communicator across two very different audiences: internal leadership who needs to understand what's happening in ML without drowning in jargon, and deployment customers who need to know what's possible and what isn't
You're comfortable working with customers directly, understanding what they're trying to accomplish, and translating that into a deployment plan
You don't wait to be asked—you proactively identify what customers need before they articulate it, and you proactively keep leadership informed without being prompted

Responsibilities

Customer deployment unblocking: Get customers from zero to a working custom model deployment on Anvil hardware without requiring Anvil engineering involvement
Research translation: Continuously scan robot learning research and off-the-shelf models; identify and implement what's most impactful for customers in the near term—with a hard ceiling of 100 days per initiative
Internal education: Keep Anvil leadership informed about what's happening in robot learning and why it matters for the business—in plain language, on a regular cadence
Voice of customer for hardware: Synthesize customer deployment requirements into actionable input for hardware roadmap decisions