Robotics Intern — Embodied AI & Mechanical Engineering at ANNGINE Inc.

Location: Santa Clara, CA | Onsite

Duration: 3–6 months, minimum 3 days per week

Eligibility: Current BS / MS / PhD students, recent graduates, and gap-year researchers all welcome.

We are hiring across two internship tracks within our robotics team. Please review both tracks below and apply to the one that best matches your background and interests. You are not expected to meet the qualifications for both tracks.

Track A — Research Intern: Robot Learning / Embodied AI

What You'll Do

• Train, fine-tune, and evaluate VLA, WAM, VLM, or Transformer-based models for embodied tasks, working from our existing codebase and infrastructure.
• Own one well-scoped research question end to end, from hypothesis through experiment design, execution, and honest interpretation of the results.
• Work with real robot data and, depending on your project, real robots — closing the loop between training and physical execution.
• Diagnose failures across data, architecture, training dynamics, and deployment, and turn what you find into measurable improvements.
  
  

• Solid Python and PyTorch, and the ability to navigate an unfamiliar training codebase without hand-holding.
• Working understanding of Transformers and modern training practice, including the mechanics that determine whether a run succeeds or falls apart.
• Evidence you've trained or fine-tuned models yourself. Scale matters far less than genuine ownership and understanding of what happened.
• Comfort with ambiguity. Research direction shifts often; we offer context and guidance rather than a ticket queue.
• Clear communication, including the ability to report a negative result plainly and ask a good question.
  
  

• Genuine obsession with embodied AI. You follow the field closely enough to hold independent views on where it's heading.
• Hands-on experience running VLA or WAM models on real robots. Beyond simulation — you've deployed policies on physical hardware and dealt with the failure modes that evaluation metrics don't capture.
• Fluency with the modern AI toolchain. AI coding tools and agent frameworks are a native part of how you work, and you have judgment about where they help and where they don't.
• Learning velocity, proven by what you've built. We'd rather work with someone who absorbs an unfamiliar method in weeks than someone who has practiced one for years, and a body of self-directed work is the clearest evidence.
  
  

• Experience with DiT-based policies, motion planning, or reinforcement learning.
• Simulation experience with MuJoCo, Isaac Sim, or Genesis, including Sim2Real.
• Familiarity with open-source VLA implementations and applied training pipelines.
• Multi-GPU or distributed training exposure, including JAX.
• Published or submitted research at relevant venues.
• Strong results in robot learning, model training competitions, or hackathons.
• Robotics middleware, hardware, or teleoperation experience, including ROS2.
• Deep understanding of robot dynamics and kinematics.
  
  

• Direct mentorship from senior researchers and engineers.
• Real compute for your experiments, and access to physical robots.
• A project with a real chance of publication, open-source release, or deployment on hardware.
• Full inclusion in the team's research and design discussions.
• Competitive intern compensation, plus a strong path to a full-time offer for interns who excel.
  
  

• Design robot mechanisms and subassemblies in CAD, and carry them through prototype, integration, and test on real hardware.
• Drive fast design iteration through additive manufacturing — print, test, revise, repeat, often inside a single day.
• Validate designs in simulation before committing to hardware, then check the model against what the bench actually does.
• Diagnose failures across structure, actuation, tolerancing, and integration, and turn what you find into measurable improvements.
  
  

• Strong parametric CAD proficiency, including assembly modeling and drawing practice.
• Real 3D printing experience — you own the loop from model to printed part, and understand how process, orientation, and material choice govern what comes off the plate.
• Working fluency with simulation, whether structural, dynamic, or full-robot, plus healthy skepticism about results that haven't been checked against hardware.
• First-principles thinking: you reason from the underlying physics rather than from precedent, and you understand how a mechanical choice propagates through actuation, control, and software.
• Enough programming ability — Python or equivalent — to script a test, process data, and work comfortably alongside the software team.
• Comfort with ambiguity and clear communication, including the ability to report a failed test plainly.
  
  

• Genuine obsession with robots. You follow how they're actually built closely enough to hold independent views on what separates a good design from a fragile one.
• Hands-on experience carrying a robot or mechanism from concept to working hardware. Beyond coursework — you've built something that moved, watched it fail in ways the model didn't predict, and fixed it.
• Fluency with the modern AI toolchain. AI coding and design tools are a native part of how you work, and you have judgment about where they help and where they don't.
• Learning velocity, proven by what you've built. We'd rather work with someone who absorbs an unfamiliar discipline in weeks than someone who has practiced one for years, and a body of self-directed work is the clearest evidence.
  
  

• Design-for-additive judgment, or experience beyond desktop FDM.
• Actuator, transmission, or compliant-mechanism design.
• Sim2Real experience, or building simulation environments rather than only using them.
• Electronics, embedded systems, or sensor integration.
• Strong results in robotics competitions or comparable team build projects.
  
  

• Direct mentorship from senior robotics and mechanical engineers.
• Real printers, fabrication resources, and hardware to build on.
• A project with a real chance of ending up on a robot that ships.
• Competitive intern compensation, plus a strong path to a full-time offer for interns who excel.
  
  

• Robot Learning / Embodied AI
• Mechanical Robotics Engineering
• Open to Either