Resume
Basics
Name | Shailesh |
Label | Undergraduate |
shailesh.xml@gmail.com | |
Summary | A highly motivated student with a passion for robotics and artificial intelligence. |
Work
-
2024.12 - Present MARMOT Lab: National University of Singapore (NUS)
Researcher Intern
Project: Long Horizon Mobile Manipulation.
- Developing a manipulation-aware locomotion policy for long-horizon mobile manipulation tasks.
- Fine-tuned generalist πο (VLA) manipulation policy on custom dataset.
- Previously worked on fine-tuning VLM models for reason based image segmentation.
-
2024.08 - 2025.05 CIR: IIIT Allahabad
Research Intern
Project: Robotic Grasping
- Designed grasping policy leveraging ViT-based visual features.
- Performed real robot experiments with Kinova Lite manipulator.
-
2024.05 - 2024.07 ERC Lab, National Taiwan Normal University
Robotics Intern
Project: Collaborative control of upper & lower body of wheeled humanoid robot using multi-agent RL.
- Trained robust policy to maneuver and jump in Isaac Gym environment.
- Trained collaborative policies for upper and lower body, using MAPPO, IPPO & Dual-A2C.
- Developed a unified policy enabling the upper body to perform manipulation tasks while the lower body concurrently handled balance and maneuvering.
-
2023.12 - 2023.12 Stochastic Robotics Lab, Indian Institute of Science (IISc, Bangalore)
Robotics Intern
Project: RL-based locomotion controller of legged robot.
- Trained a unified RL policy achieving 4 distinct gait behaviors, improving locomotion diversity.
- Robot trained on teacher-student curriculum learning.
- Developed a ROS package for the same robot platform.
- Transferred the RL policy into ROS (Gazebo) environment.
-
2023.02 - 2024.04 TEXMIN (Technology Innovation Hub)
Project Assistant
Project: Development of in-house Quadrupedal Robot for Mining Application.
- Developed ROS based framework for a in-house developed Quadrupedal Robot.
- Derived kinematics & dynamics for the robot.
- V1: Locomotion controller using ZMP stabilization.
- V2: Reinforcement Learning based control policy.
- Hardware testing of locomotion controller.
Education
-
2021.08 - 2025.05 Dhanbad, India
-
2018.04 - 2021.03 Gaya, India
Publications
-
2025 Search-TTA: A Multimodal Test-Time Adaptation Framework for Visual Search in the Wild
CoRL 2025
Submitted to CoRL 2025.
-
2025 GRIM: Task-Oriented Grasping with Conditioning on Generative Examples
RSS 2025 and ICML 2025 Workshops
Accepted at RSS 2025 and ICML 2025 Workshops.
Awards
- 2023.01.01
1st Position, Robofest Gujarat 3.0 2023
Robofest Gujarat
India's Biggest Robotics Competition, prize money $12,000
- 2022.01.01
1st Position, Vichesta (ROS simulation competition), Takshak 2022
Takshak 2022
East India's Largest Robotics Fest
- 2022.01.01
Skills
Programming | |
Python | |
C++ | |
Linux | |
Git |
ML/AI | |
RL | |
ViTs | |
VLMs | |
Diffusion Models |
Simulation/Visualization | |
Isaac Sim | |
Gazebo | |
RViz | |
CoppeliaSim | |
Simulink |
Frameworks/Libraries | |
ROS/ROS2 | |
PyTorch | |
Open3D | |
MoveIt | |
OpenCV | |
Matplotlib |
Hardware | |
Jetson Orin | |
Microcontrollers | |
Servo Actuators | |
Power Electronics | |
IMUs | |
3D Printing |
Interests
Robotics | |
Computer Vision | |
Legged Robots | |
Humanoid Robots | |
Control Systems |
Embedded Systems | |
Microcontrollers | |
Sensor Interfacing |
Machine Learning | |
Reinforcement Learning | |
Foundation Models | |
Computer Vision |
Projects
- 2024.08 - Present
mini_GPT
Personal Project. Implemented GPT-3 style transformer model architecture from scratch. Model was trained on Shakespeare Dataset.
- 2022.12 - 2023.02
Inter IIT Tech Meet 11.0 - Drona Aviation Problem Statement
Worked on accurate pose estimation of drones using ArUco Tag and a single monocular camera.
- Hovering a pluto drone on a particular height using ArUco Tag.
- Move the drone in rectangular motion (1 x 2 meter).
- Make the second drone follow the first drone.
- 2023.03 - 2023.12
Two-Wheeled Self Balancing Robot
Developed a custom cascaded PID-based controller for a two-wheeled self-balancing robot.
- Developed a custom cascaded PID based controller that incorporate acceleration value as well.
- Simulated our robot in Gazebo(cascaded PID controller) and in Simulink(with LQR controller)
- Designed and build a two wheeled robot in hardware from scratch.