Bhavesh Bakshi
Contact

Bhavesh
Bakshi

Robotics Engineer — RL, Control, Systems

Bhavesh Bakshi

Real-Time Control

50Hz PPO-based inverse kinematics within a 500Hz safety orchestrator, deployed on real humanoid hardware.

PID Controller Playground INTERACTIVE

A ball chases your cursor using a PID controller. Tune Kp, Ki, Kd to see oscillation vs smooth tracking.

Learning Systems

Training locomotion and manipulation policies across thousands of parallel simulations, deploying them on real robots.

Cart-Pole Balancer LIVE TRAINING

Watch a neural network learn to balance a pole in real-time using REINFORCE. Episodes improve as the policy gradient converges.

Q-Learning Gridworld PLAYABLE

Click to place the agent. Watch it learn to avoid lava and reach the goal via Q-learning. Cells show learned values.

Maze Solver RACE

A*, BFS, and DFS race to solve a generated maze. Compare explored nodes, path length, and speed across algorithms.

Human-Robot Interaction

XR teleoperation, dexterous hand retargeting, and sim-to-real deployment pipelines for humanoid control.

2D Robot Navigator INTERACTIVE

Click to place obstacles, set start and goal. Watch RRT explore the space and find a collision-free path in real-time.

Projectile Optimizer GRADIENT DESCENT

Click to set a target. Gradient descent optimizes launch angle and velocity to hit it. Watch the loss landscape converge in real-time.

N-Body Gravity SIMULATION

Click to spawn masses. Bodies interact via gravitational attraction using Velocity Verlet integration. Drag to set initial velocity.

INTERACTIVE LAB

Want more?

63 interactive simulations — physics, AI/ML, control theory, signal processing. All running in your browser.

Explore the Lab →

Hardware & Build

From foam-board aircraft to FEA-optimized chassis. Hands-on fabrication, flight testing, and mechanical design.

Bhavesh Bakshi

About

Robotics engineer who builds systems that move in the real world. I work at the intersection of reinforcement learning, real-time control, and hardware deployment. Currently at SS Innovations, building real-time control systems and deploying RL-trained motion policies on robotic platforms.

Stack
Languages: Python, C++, Bash
RL/Sim: Isaac Gym, Isaac Lab, MuJoCo, PPO, SAC
Robotics: ROS2, CycloneDDS, Pinocchio, ONNX Runtime
Hardware: Unitree G1/H1, Quest 3, RealSense, Arduino
Tools: PyTorch, SolidWorks, Ansys, Docker, Git

Experience

Mechatronics Engineer SS Innovations Pvt. Ltd.
Oct 2022 — Present · Gurugram, India
  • Building RL-trained motion policies and real-time arm control systems (50Hz policy, 500Hz safety loop) for humanoid robotic platforms
  • Implemented XR teleoperation with Quest 3 for dual-arm manipulation at <50ms LAN latency
  • Led end-to-end integration and testing of the SSI Mantra surgical robotic system as System Integration Incharge
  • Automated harness testing for instrument actuators and active arm calibration workflows
  • Tuned and validated active arm performance across manufacturing test cycles
Python C++ ROS2 Isaac Gym PyTorch MuJoCo Test Automation System Integration
Mechanical Design Intern Learning Links Foundation 2021
Mechanical Intern Heavy Engineering Corporation 2020

B.E. Mechanical Engineering — BIT Mesra, Ranchi (2018–2022)

Get In Touch

Available for robotics roles