About.
I like problems where the software has to touch something real.
I did my M.Sc. in Robotics and Embedded AI at Maynooth. Passed in November 2025, graduated in March 2026. My thesis pointed a camera at oat and barley plots and worked out how stressed the plants were before anything showed up to the eye. It got leaf temperature to about 2.1 °C. Before Ireland I studied Mechatronics in Chennai.
The three that actually taught me something were the hard ones. Fusing RGB and multispectral streams so the network could read a physiological signal it has no business seeing from one static frame. Turning a FarmBot CAD export of 1,174 parts into five links Gazebo would simulate and MoveIt could plan against. And writing a browser agent that knows when its own click did nothing, which is a much harder question than it sounds.
None of that was the fun part. It was weeks of shrinking meshes, and a few days lost to an ESP8266 that kept resetting before I found the power rail sagging under the motor driver. I don't really mind. I'd rather hand someone something that runs on real hardware than a notebook that worked once on my laptop.
What I use, grouped by discipline.
- →Deep Learning (TensorFlow / Keras)
- →Computer Vision
- →Reinforcement Learning
- →PyTorch
- →TensorFlow.js
- →Multimodal / FiLM fusion
- →ROS 2 · Gazebo · MoveIt 2
- →URDF / Xacro modelling
- →Arduino (Mega / Uno)
- →ESP8266 / ESP32
- →Sensor integration
- →Motor control (L298N / servos)
- →Python
- →TypeScript / React
- →Chrome Extensions (MV3)
- →Unity / AR Foundation
- →Flask / REST APIs
- →Git · Linux
Timeline.
- 2024 to 2025M.Sc. Robotics & Embedded AIMaynooth University, Ireland
Thesis: Multispectral analysis of plant physiological stress. Passed November 2025, graduated March 2026.
- 2019 to 2023B.Tech. Mechatronics EngineeringHindustan Institute of Technology & Science
Autonomous warehouse robot · bionic arm capstone.