Closing the loop between code and the physical world.
Robotics & embedded systems engineer
I build the software that lets machines sense, decide and move.
Robots should be practical enough to use, reliable enough to trust, and safe enough to work beside people.
That standard is set where software stops being abstract: in motor currents, sensor noise, bus timing and safety limits. My work is to carry ideas across that line and make them hold up in the real world, so that robotics becomes more accessible and more useful to the people it works with.
ROS 2 architectures that turn intent into coordinated motion, from gripper control to navigation and human-robot interaction.
Firmware in C and C++ on ESP32-S3 and STM32: motor control, sensor feedback and the buses that hold a system together.
Python for data and signal processing, TypeScript and React for interfaces: the tooling that makes hardware observable and usable.
I studied Electronics and Communication Engineering at Saranathan College of Engineering. Circuits, embedded systems and connected devices gave me the fundamentals. An IoT internship and a smart agriculture system built for the Smart India Hackathon gave me a first taste of building things that sense the world.
What happens at the point where software meets the physical world? Following that question led me to robotics and intelligent automation.
I moved to Germany for a Master's in Electrical Engineering at the University of Stuttgart, focusing on robotics and embedded systems.
I joined Pilz GmbH & Co. KG as a software developer, building frontend components in React and TypeScript. At the university I built the interface of the SynTrac Ergometer App, and designed a flow-controlled experiment on slide electrification, with Python doing the signal processing and statistics.
My Master's thesis: a ROS 2 software architecture for an intelligent gripper in construction robotics. Three layers, from a ROS topic on the host down to the motor, tested with 5 million commands.
As a robotics and functional safety intern I researched and designed AI-powered humanoid robot applications for industrial automation, bringing together machine vision, robot navigation and functional safety for real manufacturing scenarios.
I focus on intelligent robotics, embedded software and industrial automation. I like problems where several disciplines meet, and I am happiest when I understand a system from the pins up to the behaviour.
Analytical, adaptable, and comfortable on both sides of the hardware-software line. Select a pin to see where I have used it.
Pin description
ROS 2 Robotics
A ROS node takes commands on /gripper_cmd and publishes force, position, speed, torque and status back to ROS.
Used in: A ROS 2 software architecture for an intelligent gripperMaster's thesis, 2026. University of Stuttgart, IAS with IWB Construction Robotics
Construction robots handle heavy, irregular building materials, and the gripper is where they meet the world. I built the software that connects a ROS 2 application to the gripper's embedded controllers, then measured how well the link holds under load.
A ROS node takes commands on /gripper_cmd: open, close, stop, home, set speed, position or torque. A gripper driver packs each one into a compact binary packet.
Carries commands from the host to the gripper and feedback back, reliably and in order.
Receives packets over TCP and forwards them to the motor controller. A protocol abstraction interface puts UART and CAN behind one API, so the bus can be swapped with minimal change. It also reads the force sensor.
Both links were implemented between gateway and motor controller, so they could be compared on the same hardware.
Firmware on an STM32G431 executes the motor action and returns acknowledgement and status. Back on the host, feedback is published to ROS as force, position, speed, torque, status, acknowledgement and error topics.
Across 5 million command transmissions, CAN held above 99.9% reliability with almost no timeouts. UART was faster, but its reliability fell as the command rate rose. CAN was selected.
| Protocol | 250 Hz | 350 Hz | 400 Hz | 450 Hz |
|---|---|---|---|---|
| CAN | >99.9% | >99.9% | >99.9% | >99.9% |
| UART | ≈36% | ≈19% | ≈9% | ≈5% |
ROS 2, C/C++, ESP32-S3, STM32G431, CAN, UART, Ethernet (TCP)
Researched and designed AI-powered humanoid robot applications for industrial automation, bringing machine vision, robot navigation and functional safety together for real manufacturing scenarios.
Humanoid robotics, machine vision, navigation, functional safety
Designed a flow-controlled experiment to study how droplet velocity affects slide electrification, measured with a flowmeter, oscilloscope and current amplifier, and analysed in Python.
Python, instrumentation, signal processing, statistics
Developed the user interface of the SynTrac Ergometer App in JavaScript and improved its usability and frontend functionality.
JavaScript, UI development
Built frontend components with HTML, CSS, JavaScript, TypeScript and React, with a focus on usability, performance and maintainable code.
React, TypeScript, JavaScript, HTML, CSS
Applied supervised learning and remote-sensing methods to Landsat imagery to detect changes in water, urban and forest cover, using image segmentation and NDVI.
Supervised learning, remote sensing, NDVI
A smart-agriculture concept that estimates soil-fertility conditions in real time from soil and air moisture sensors on a NodeMCU, with an app that shows the readings and recommendations.
Sensors, NodeMCU, IoT
I am open to roles, research collaborations and interesting engineering projects in robotics, embedded software and automation.