Operations Director

Job highlights Identified by Google from the original job post Qualifications • Education: MS/PhD or equivalent deep experience in Robotics or Embodied AI • Technical Expertise: Strong background in Imitation Learning, RL, Diffusion Policies, or VLA models • ML Tooling: Proficiency in PyTorch and modern large-scale model training workflows • Systems Thinking: Deep understanding of the intersection between perception, policy, and control • Real-World Experience: Proven experience with physical robots (not simulation only) • Mindset: High ownership mindset with the ability to thrive in ambiguity and fast-paced iteration cycles • Authorization: U.S. work authorization required (Visa transfer supported) • You have successfully trained and deployed learned policies onto physical robotic systems • You have experience debugging "messy" real-world hardware and software failures • You enjoy being close to hardware and have designed your own experimental frameworks • You are driven to build and scale systems rather than just optimizing benchmarks • Build the Data Engine: Help define the data scaling laws for embodied intelligence from the ground up • 9 more items(s) Benefits • COMMITMENT Full-time / Founding Team • COMPENSATION Competitive Salary + Meaningful Equity, Pay Frequency: Monthly Responsibilities • As a Founding Robotics Researcher, you will own the VLA and policy learning direction • You won't just consume datasets; you will define the data strategy, ship models onto real robots, and design experiments that directly improve deployment performance • This role is for someone who wants to build a research engine inside a startup and is comfortable switching from PyTorch to hardware debugging when necessary • VLA & Policy Training: Architect and train VLA models for real-world tasks and design fine-tuning pipelines using deployment-collected data • Data System Design: Develop teleoperation data collection frameworks and build filtering, curation, and scaling pipelines to address distribution gaps • Hardware Integration: Deploy policies to physical robot arms and sensor stacks, tuning latency, calibration, and system reliability • Research–Deployment Loop: Build internal benchmarks tied to actual tasks and translate model failures into data and system improvements • Systems Debugging: Hands-on work with robot arms, grippers, and multi-camera systems to debug perception, policy, and control loops • Experimental Leadership: Define the experiments that matter and develop evaluation metrics tied to physical success rates • 6 more items(s) More job highlights Job description FOUNDING ROBOTICS RESEARCHER

JOB TITLE Founding Robotics Researcher

LOCATION Onsite (Bay Area, CA)

COMMITMENT Full-time / Founding Team

COMPENSATION Competitive Salary + Meaningful Equity, Pay Frequency: Monthly

ABOUT DEEPREACH

DeepReach is building the infrastructure layer for real-world embodied AI. We focus on large-scale teleoperation data, Vision-Language-Action (VLA) training, and real... deployment environments—not staged demos.

We are not a simulation-only research lab; we train on real distributions, deploy in real environments, and iterate fast to close the loop between model training and physical performance.

THE ROLE

As a Founding Robotics Researcher, you will own the VLA and policy learning direction.

You won't just consume datasets; you will define the data strategy, ship models onto real robots, and design experiments that directly improve deployment performance.

This role is for someone who wants to build a research engine inside a startup and is comfortable switching from PyTorch to hardware debugging when necessary.

KEY RESPONSIBILITIES

• VLA & Policy Training: Architect and train VLA models for real-world tasks and design fine-tuning pipelines using deployment-collected data.

• Data System Design: Develop teleoperation data collection frameworks and build filtering, curation, and scaling pipelines to address distribution gaps.

• Hardware Integration: Deploy policies to physical robot arms and sensor stacks, tuning latency, calibration, and system reliability.

• Research–Deployment Loop: Build internal benchmarks tied to actual tasks and translate model failures into data and system improvements.

• Systems Debugging: Hands-on work with robot arms, grippers, and multi-camera systems to debug perception, policy, and control loops.

• Experimental Leadership: Define the experiments that matter and develop evaluation metrics tied to physical success rates.

REQUIREMENTS

• Education: MS/PhD or equivalent deep experience in Robotics or Embodied AI.

• Technical Expertise: Strong background in Imitation Learning, RL, Diffusion Policies, or VLA models.

• ML Tooling: Proficiency in PyTorch and modern large-scale model training workflows.

• Systems Thinking: Deep understanding of the intersection between perception, policy, and control.

• Real-World Experience: Proven experience with physical robots (not simulation only).

• Mindset: High ownership mindset with the ability to thrive in ambiguity and fast-paced iteration cycles.

• Authorization: U.S. work authorization required (Visa transfer supported).

STRONG SIGNALS

• You have successfully trained and deployed learned policies onto physical robotic systems.

• You have experience debugging "messy" real-world hardware and software failures.

• You enjoy being close to hardware and have designed your own experimental frameworks.

• You are driven to build and scale systems rather than just optimizing benchmarks.

WHY JOIN US

• Real-World Impact: Unlike traditional labs that stop at publication, we measure success by real-world deployment performance.

• Build the Data Engine: Help define the data scaling laws for embodied intelligence from the ground up.

• Founding Ownership: Gain meaningful equity and direct influence over the company's research direction.

• Physical AI Frontier: Work at the hardest unsolved problem in robotics—making robots work reliably in production environments.

• Global AI Collaboration: Leverage Talex.ai’s ecosystem to integrate cultural and linguistic intelligence into physical systems. Show full description Report this listing Loading...