SHREYAS SHARMA
Robotics · Embedded Systems · Applied ML
B.Tech in Automation & Robotics at SRM Institute of Science and Technology, Chennai. I build autonomous systems, embedded controllers, and ML pipelines — from bare-metal firmware to full-stack dashboards. Previously an AI Engineer Intern at QID.
About
I'm Shreyas Sharma, a final-year B.Tech student in Automation & Robotics at SRM Institute of Science and Technology, Chennai.
I work across the stack — embedded firmware on ESP32 and Arduino, perception pipelines with YOLO and OpenCV, motion control with ROS2, and local backend systems with Node.js and React. My projects range from autonomous robotic arms and drone tracking systems to real-time IoT monitoring dashboards.
In summer 2025, I interned at QID as an AI Engineer, building real-time detection pipelines with YOLOv8, OpenCV, and ONNX Runtime — deployed across local systems and GCP.
I'm looking for opportunities in robotics, embedded systems, computer vision, and applied ML — anywhere I can build systems that work in the real world.
Experience
Stack
Tools and technologies I actively build with.
Core Languages
Robotics & Embedded
ML & Computer Vision
Systems & Workflow
Featured Projects
Larger systems where I own the full build — from hardware to software.
Robotic arm that autonomously locates and docks an EV charging connector using computer vision and depth sensing.
End-to-end perception-to-control pipeline. Custom Transformer-Enhanced YOLO (EVS-YOLO) for charging port detection. RGB-D camera and IMU integration for pose estimation. ROS2 nodes for frame transforms and control. Designed around a custom 5-DOF arm.
Full-stack local IoT system for real-time energy monitoring with ESP32 nodes, MQTT broker, and a React dashboard.
ESP32 firmware publishes sensor telemetry over MQTT to a local Mosquitto broker. Node.js/TypeScript backend bridges data to a React/Vite frontend via WebSocket. Real-time per-appliance visualization. Fully local — no cloud dependencies. Earlier versions used Unity 3D for digital twin visualization.
YOLO-based drone detection with ByteTrack multi-object tracking on a custom dataset.
Per-frame YOLO detection with ByteTrack for persistent object IDs across video. Custom drone dataset. Full training, inference, and evaluation pipeline. Developed and benchmarked on Apple Silicon.
Custom robotic arm with ESP32 firmware, PCA9685 servo driver, and Python-based inverse kinematics.
4× MG996R torque servos + SG90 gripper. ESP32 + PCA9685 for synchronized PWM control. Task-structured embedded firmware. Python IK solver for pick-and-place paths. Designed to integrate with the EV charging perception stack.
More Projects
Focused builds across embedded control, IoT, applied ML, and industrial automation.
Collaborative research project. Autonomous object identification and sorting using ROS2 and OpenCV. Sensor data stream integration and calibration workflows for a 6-DOF manipulator.
End-to-end classification pipeline on real crime data. Full ETL and feature engineering. Multi-classifier comparison (including XGBoost) with metrics and confusion matrices. Reproducible experiment scripts and reusable inference pipeline.
Terrain-capable six-legged walking robot. Multi-servo coordinated locomotion with gait pattern programming. Demonstrates legged robotics and multi-actuator control fundamentals.
Closed-loop control system that balances a ball on a platform using PID feedback. Real-time sensor feedback with PID tuning for stable positioning. Classical control theory applied to physical hardware.
Embedded closed-loop environmental controller. ESP32 as control master, Arduino Uno as display slave over I2C. Hysteresis-based actuator control for humidifier and exhaust fan. Real-time sensor visualization.
ESP8266-based queue management. RC522 RFID for patient verification. Dual IR sensors for entry/exit counting. Auto token generation and queue logic. Web-based frontend for real-time monitoring.
Temperature monitoring and cooling control. LM35 temperature sensing. 24V DC cooling fan controlled via relay module. PLC-compatible interface with LED pilot lights.
Experiments
Smaller explorations and learning projects.
ESP32-based soil moisture monitoring with dual pumps, auto/manual modes, and MQTT dashboard control.
Automated engagement bot with lightweight NLP-based content filtering.
Reinforcement learning experiment training an agent to play Flappy Bird.
Contact
Get in Touch
Open to internships, research collaborations, and project discussions in robotics, embedded systems, and applied ML.