Development of a six-legged robotic platform that combines nonlinear control and learning-based adaptive policies for terrain-aware locomotion.

• Modelled full kinematics and dynamics of the hexapod using URDF.
• Implemented and evaluated controllers in a ROS-based simulation environment.

CNN-based programming-by-demonstration framework that uses vision and sensor data to generalise robot motions from human demonstrations.

• Applied convolutional neural networks to recognise demonstrated tasks.
• Studied task perception, representation, and planning to enable autonomous execution of demonstrated skills.

Design of an integrated workflow for robotic-assisted bone tumour resection using high-speed drilling and laser-based soft tissue cutting.

• Implemented a tumour-fitting algorithm using BFGS optimisation to approximate tumour geometry.
• Generated 3D cutting trajectories and adaptive actuation control for precise, conformal resections.

Coursework and mini-projects within the MPhil Scientific Computing programme, focusing on parallelisation for high-performance computing and programming with GPUs (CUDA).

• Implemented parallel algorithms using MPI and OpenMP for distributed and shared-memory architectures.
• Developed CUDA kernels and analysed performance, memory hierarchy usage, and scaling behaviour.

Design and analysis of feedback control systems using MATLAB/Simulink for a range of dynamic scenarios.

• Constructed models to analyse stability, steady-state error, and settling time under different gains and inputs.
• Explored potential deployment paths for embedded platforms, including FPGA-based implementations.

AI-assisted TCAD workflow that integrates retrieval-augmented generation and program synthesis to automate PSIM-based semiconductor device simulations.

• Designed a hybrid dense–sparse retrieval pipeline (Qdrant + Ollama embeddings + BM25) for TCAD documentation and queries.
• Implemented natural-language-driven PSIM script generation with self-correction loops, improving both speed and robustness of device simulations.

Electromyography-based system for measuring muscle activity and using EMG signals to control external outputs via a microcontroller.

• Designed analogue signal conditioning with instrumentation amplification, bandpass filtering, rectification, and integration stages.
• Implemented analogue-to-digital conversion and created audio/visual displays to represent muscle activity and strength.

FPGA-based system for descrambling audio messages, combining digital filtering and mixed-signal design.

• Integrated analogue front-end circuits with a digital system designed in Quartus Prime.
• Developed a custom digital notch filter and used MATLAB for analysis and validation.

Electrical design for an industrial robotic system used in battery dismantling, including schematic design and integration with mechanical subsystems.

• Produced electrical drawings in EPLAN Electric P8 and collaborated closely with mechanical design using SolidWorks CAD.
• Refined the robotic system structure with additional actuation to improve loading and prevent cargo stacking.

Design and implementation of circuits and control algorithms for a robot capable of completing a maze using sensor feedback.

• Built subsystems enabling two-dimensional motion and obstacle detection.
• Developed and iteratively refined Arduino-based algorithms for efficient and robust maze navigation.