Background
About
I started out in information and communication technology, which I studied for my B.Tech. That is where my interest in robotics first took shape. From there I kept drifting toward the parts of engineering where software has to move something real. That pull took me into control theory and robotics research at the Indian Institute of Science, and then to Saarbrücken for a master’s in computer science and robotics research at ZeMA.
The question behind all of it is simple. How does a plan made in software hold up once a real robot has to carry it out? It has to deal with real geometry, noise, and hardware limits. Most of my work sits on that line, between an algorithm and a machine that has to act in the world.
Below is the same path, step by step.
-
Robotics research
ZeMA gGmbH · Saarbrücken
Robotics research at ZeMA
At ZeMA, I work on robots that assemble products and pick parts from cluttered bins. The pipeline runs from planning the order of assembly steps, to finding collision-free paths, to verifying everything in a digital twin before a real arm executes it. My focus is the digital-twin integration and the planning code that ties these stages together.
So far the work has led to three papers: a live assembly-picking system [1] and a collision-free trajectory planner [2] from 2025, and FAST [3], a graph transformer that predicts whether a disassembly step is physically feasible, so the planner can learn from failed attempts.
-
MSc in Computer Science
Saarland University · Saarbrücken
MSc in Computer Science
At Saarland University [1], I moved from control and robotics into machine learning. Coursework and small research projects let me work hands-on across methods: reinforcement learning from fixed rules to deep Q-networks, graph transformers, explaining model decisions with integrated gradients, classical image processing, generative models for synthetic-to-real images, and efficient adaptation of a multilingual language model. They are written up in the project index [2].
My master’s thesis is based at ZeMA, on the assembly-planning work above.
-
Research Assistant
RBCCPS, Indian Institute of Science · Bangalore
Research Assistant at RBCCPS
Back at the Robert Bosch Centre for Cyber-Physical Systems, I worked on symbolic control, a way of designing controllers that are mathematically guaranteed to stay safe. One project was safe autonomous exploration for ground robots, pairing target detection with a symbolic controller so the robot stays safe while approaching a target [1]; another was distributed controller synthesis for multi-agent systems [2].
-
Project Intern
RBCCPS, Indian Institute of Science · Bangalore
Project Intern at RBCCPS
This started with an internship under Professor Pushpak Jagtap. I modeled the dynamics of unknown systems with Gaussian processes, training on small datasets by combining kernel functions for better generalization, which became my first paper [1]. Alongside that, I built ROS support for a PhaseSpace motion-capture system and control algorithms for robots running under complex specifications.
-
BTech in Information and Communication Technology
Pandit Deendayal Energy University · Gandhinagar
BTech in Information and Communication Technology
My undergraduate years at Pandit Deendayal Energy University are where robotics and machine learning first caught hold. As co-president of the Cretus robotics club, I helped build a chess-playing robotic arm, an early lesson in how fragile a “working” demo gets once lighting and timing change. During the lockdown, I taught myself deep learning and built a face-recognition system on a convolutional network written from scratch, deployed for identity verification from live video [1]. I also took part in the e-Yantra robotics competition [2] and open-source challenges for Google Summer of Code [3], alongside course projects in AI and embedded systems.