Robotics Engineer

Miko; Mumbai; On-site; INR 900,000 - 1,300,000; Full time

Job summary

Miko in Mumbai is seeking a robotics engineer to join a compact, technically dense team. You will work on learning-based control for diverse robotic systems, deploying to physical robots and developing sim-to-real capabilities.

Responsibilities include modeling manipulators, building simulators, hardware bring-up, and integrating classical control with learned policies. Strong foundation in robotics and programming is essential; Linux/Git proficiency is expected.

Qualifications

• 1–3 years of hands-on robotics experience (industry, research lab, or strong open-source work).
• Strong grounding in classical robotics: forward/inverse kinematics, rigid-body dynamics, trajectory generation, feedback control, transforms.
• Experience with robot modeling (URDF/MJCF/SDF) and at least one physics simulator (MuJoCo, Isaac Gym/Gym, Gazebo, PyBullet).
• Proficient in Python and C++.
• Comfortable on Linux, with Git, and working close to hardware.
• Clear communicator who can reason from first principles and validate claims against real data or source code.

Responsibilities

• Model and characterize robot manipulators: kinematics, dynamics, coordinate frames, calibration.
• Build and maintain robot models and simulation environments, ensure sim behavior matches real arms.
• Bring up hardware: servo/motor control, sensor integration, teleoperation and data-collection rigs.
• Bridge classical control and learned policies to ensure safe, reliable motion on hardware.
• Create data flywheel for model fine-tuning for multiple tasks and embodiments.
• Set up evaluation: reproducible benchmarks in sim and on the robot, with honest metrics.
• Debug real-world failures that don't appear in papers.

Skills

Robotics experience

Classical robotics

Python

C++

Linux

Git

Strong communicator

Tools

URDF

MJCF

SDF

MuJoCo

Gazebo

PyBullet

Isaac Sim

OpenAI Gym

Job description

We're building learning-based control for diverse robotic systems, aimed at real-world deployment in industrial setups. We pair classical robotics - kinematics, dynamics, perception, navigation, control - with modern vision-language-action (VLA) models, and we care as much about what happens on real hardware as in simulation.

You'll join a small, technically dense team where your work ships to physical robots quickly. If you like the space between "the model said to move here" and "the arm actually moved there, safely," this is that job.

What you'll do

• Model and characterize robot manipulators: kinematics, dynamics, coordinate frames, calibration.
• Build and maintain robot models and simulation environments, and keep sim behavior faithful to the real arms.
• Bring up hardware: servo/motor control, sensor integration, teleoperation and data-collection rigs.
• Work at the boundary between classical control and learned policies - turning a policy's output into safe, reliable motion on hardware.
• Creation of data flywheel for model fine-tuning for multiple tasks and embodiments
• Set up evaluation: reproducible benchmarks in sim and on the robot, with honest metrics.
• Debug the messy real-world failures that never show up in a paper.

What we're looking for

• 1–3 years of hands-on robotics experience (industry, research lab, or a serious body of personal/open-source work).
• Strong grounding in classical robotics : forward/inverse kinematics, rigid-body dynamics, trajectory generation, feedback control, transforms.
• Experience with robot modeling (e.g. URDF / MJCF / SDF) and at least one physics simulator (MuJoCo, Isaac Sim/Gym, Gazebo, PyBullet, or similar).
• Proficient in Python and C++ .
• Comfortable on Linux, with Git, and working close to hardware.
• Clear communicator who can reason from first principles and validate claims against real data or source code rather than intuition alone.

Nice to have (a plus, not required)

• A portfolio of robotics work - repos, demos, videos, publications, competition entries, or a personal project we can look at.
• Exposure to learning-based robotics: imitation learning, VLA models, or RL for control.
• Hands-on time with real manipulators and grippers.
• Background in optimization or optimal control.
• Familiarity with the LeRobot / OpenVLA ecosystem.
• Real robots, real deployment targets, short loop from idea to hardware.
• A team strong in controls, safety, and systems - you'll learn fast.