Select a drone to fly · WASD move / QE yaw / RF height · Drag to orbit · Esc exits
Software systems, data analysis, and robotics. Computer Science PhD researcher at Universiti Malaya.
A closer look at the systems I build, study and help bring together.
Mechanical engineering taught me to connect a model with the machine in front of me: its tolerances, loads, and failure modes.
R&D coordination brought suppliers, design, technical teams, and production into the same conversation. I learned to make requirements clear enough for someone else to act on.
ROS, Linux, Python, and C++ became everyday tools. Setting up simulations and tracing runtime problems brought software into the centre of my work.
My current work at Universiti Malaya brings together UAV autonomy, swarm research, and AI-assisted software development. Each experiment gives me another reason to revisit the system.
Autonomous systems that can coordinate, adapt, and be tested.
How can local decisions produce intelligent swarm behaviour?
I am interested in how UAVs coordinate with limited communication and partial information. My research questions connect distributed decision-making, formation behaviour, and the constraints of local trajectory planning.
From a research question to a repeatable experiment.
I use the open-source EGO-Planner implementation as a baseline for my UAV research. ROS, Gazebo, and RViz help me set up scenarios, inspect trajectories, and compare behaviour as I change the assumptions.
What survives the move from simulation to hardware?
Sensing uncertainty, latency, compute limits, and hardware variation shape the questions I want to test. I use logs and repeatable checks to trace failures and decide what the next experiment should examine.
I define the task. Agents help me build and investigate. The results guide the next decision.
I set the requirements, choose the trade-offs, and review the outcome.
A task an agent can act on.
I provide the relevant files, constraints, expected behaviour, and a clear stopping point. Independent tasks can run in parallel; shared decisions stay connected.
Code, experiments, and a record of changes.
I use coding agents to implement changes, inspect failures, and draft technical notes. Git keeps the changes inspectable so I can review what was actually produced.
The next step comes from the result.
I check the behaviour against the original task, read the test output, and look at the interface or experiment. Findings return to the task brief; an accepted result gets documented for reuse.
Versioned code · Test results · Decisions · Handoff notes
Another way into the projects: through a setup, a design choice, or an observation from the work.
I welcome conversations about robotics, software tools, and research. If you have a project in mind, I would be glad to hear about it.