Robot Learning Engineer Intern

Gritt; South San Francisco (CA); On-site; USD 48,000 - 69,000; Part time

Benefits offered by this job

Mentorship

Opportunity for return or full-time.

Job summary

Gritt, a robotics and AI company, seeks a Robot Learning Engineer Intern in the SF Bay Area (in-person) for 3 or 6 months. You own a defined deliverable, work with a dedicated mentor, and demo your work to the team. Internships offer competitive salaries and a mission-driven climate impact.

You will train policies for real robot tasks, build data pipelines, deploy on hardware, and test on real robots. Applicants should be pursuing MS/PhD in ML/Robotics/CS or related fields and have strong

Qualifications

• Pursuing MS/PhD in ML, Robotics, CS, or related field (exceptional undergraduates welcome).
• Strong PyTorch; solid grounding in deep learning and one of IL/RL.
• Applied evidence: publications, open-source, or substantial course/competition projects.
• Should be comfortable taking ownership of tasks with light supervision.
• Must have excellent problem-solving skills.
• Legally authorized to do an internship in the United States for either 3 months or 6 months.

Responsibilities

• Train policies (imitation learning, RL, or fine-tuned foundation models) for real robot tasks.
• Build data collection and evaluation pipelines; run ablations.
• Deploy policies on hardware and iterate against real-world performance.
• Test your work on real robots at the office.

Skills

PyTorch

Deep learning

Imitation Learning

Reinforcement Learning

Ownership mindset

Problem solving

Education

MS/PhD in ML, Robotics, CS, or related field

Tools

Git

Job description

Gritt is an intelligent system that combines robotics and AI to build the infrastructure that pulls society forward. Gritt deploys via simple attachments to common equipment found on construction sites and autonomously performs labor-intensive tasks, verification, and planning. Gritt systems are already building critical infrastructure in the harshest outdoor environments, starting with large-scale solar. The founding team includes experts in robotics and AI from Carnegie Mellon, Stanford, and MIT. Gritt is backed by Obvious Ventures, Union Square Ventures, First Round Capital, Climactic, Congruent Ventures, and other leading firms.

Role: Robot Learning Engineer Intern

Location: SF Bay Area (in-person)

About Internships at Gritt

Our internships are scoped projects: you own a defined deliverable end-to-end, work with a dedicated mentor, and demo your work to the whole team. Many interns receive return or full-time offers. This will be an internship for one of two durations: 3 months, or 6 months.

We offer competitive salaries, and the opportunity to work on a mission with tremendous climate impact.

What you'll get to work on

• Train policies (imitation learning, RL, or fine-tuned foundation models) for real robot tasks.

• Build data collection and evaluation pipelines; run ablations.

• Deploy policies on hardware and iterate against real-world performance.

• Test your work on real robots at the office.

• Get hands-on with pose estimation, VLM, VLA, 3D scene reconstruction, object detection, and multimodal learning.

• Get the opportunity to publish (for PhD interns).

• Attend Tier-1 industry conferences.

An example project could be could be anything from pick-and-place policies for large flexible panels under wind disturbance, to sim-to-real for manipulation from a machine on uneven ground.

What we look for

• Pursuing MS/PhD in ML, Robotics, CS, or related field (exceptional undergraduates welcome).

• Strong PyTorch; solid grounding in deep learning and one of IL/RL.

• Applied evidence: publications, open-source, or substantial course/competition projects.

• Should be comfortable taking ownership of tasks with light supervision.

• Must have excellent problem-solving skills.

• Legally authorized to do an internship in the United States for either 3 months or 6 months.

Nice to have

• Experience with simulators (Isaac, MuJoCo), VLA/foundation models, tele-op data collection, or real-hardware deployment.