AI/ML & Robotics Engineer in Progress | Deep Reinforcement Learning | Applied AI
B.Tech – Mathematics & Computing
- Mathematics & Computing undergraduate focused on AI, Machine Learning and Robotics
- Building intelligent systems using Deep Reinforcement Learning
- Experienced with PPO, SAC, hierarchical control and classical feedback controllers
- Interested in combining learning-based control with physics-based simulation
- Currently focused on Robotics, Reinforcement Learning and Applied AI
IEEE ICC 2027 - Under Review
- First-author research work investigating fault-tolerant hover control for the Crazyflie nano-quadrotor.
- Benchmarks PID, SAC and Domain-Randomized SAC under progressive actuator degradation.
ACODS 2026 - Accepted
- Co-authored research on a ROS 2 framework for resilient quadrotor swarms.
- Combines firmware-level fault injection with dynamic topology reconfiguration.
| Area | Tools & Concepts |
|---|---|
| Reinforcement Learning | PPO, SAC, Stable-Baselines3, Gymnasium, Reward Shaping, Curriculum Learning, Domain Randomization, Fault Injection |
| Robotics & Control | MuJoCo, PyBullet, ROS2, Manipulation, Quadruped Locomotion, Quadrotor Control, Kinematics, PID/PD, Torque Control, Fault-Tolerant Control |
| ML & Deep Learning | PyTorch, TensorFlow, Scikit-Learn, OpenCV, Neural Networks, Computer Vision, Model Evaluation |
| Programming & Data | Python, C++, TypeScript, NumPy, Pandas |
| Generative AI & Retrieval | FastAPI, React, BM25, Groq, RAG, LLM Integration, Grounded Generation |
| Development & Deployment | Git, Streamlit, Vercel, Render, REST APIs, Model Deployment |
- Vision-free pick-and-place control for a 7-DOF Franka Panda
- Hierarchical PPO + PID control in MuJoCo
- Compared learning-based control against classical IK + PID
- PPO-based locomotion controller for Unitree A1
- Joint-level PD torque control with PyBullet
- Extended to uneven terrain using curriculum learning
- Goal-conditioned SAC for actuator fault-tolerant hover control
- Custom Gymnasium + PyBullet environment with motor fault injection
- Domain-randomized control evaluated against PID and SAC baselines
- Vector-database-free document Q&A using BM25 retrieval
- Section-aware chunking and deterministic reranking
- React + FastAPI + Groq deployment with source citations
Deep Reinforcement Learning · Robotics · Robot Control · Autonomous Systems
Quadruped Locomotion · Robot Manipulation · Quadrotor Control · Fault-Tolerant Robotics
Applied Machine Learning · Generative AI · Retrieval-Augmented Generation
Looking for opportunities in:
Deep Reinforcement Learning · Robotics · AI/ML Engineering · Autonomous Systems · Robotics Software
Open to internships, research roles, and applied engineering opportunities.
