If thought is information — and a brain–computer interface can decode it — then what is a mind made of?
That question isn't a research project. It's why I chose this field. The work that gets me there is concrete: signal processing, statistical learning, and computational models of how neural circuits actually compute.
Signals decoded
Wi-Fi CSI Fall Detection
An ESP32 reading Channel State Information to detect falls without a camera or a wearable — privacy-preserving monitoring for elderly care. 3rd Place, Healthcare IoT Innovation & Solution Challenge, ICT Cup 2026.
Breath Bright
An AI respiratory-health screening app built to counter youth vaping, funded by ThaiHealth and piloted in real schools alongside the Smoke Free+ campaign. Technical Lead — the sole engineer in a 3-person founding team. 1st Place, Thailand National Championship — UBI-Alpha 2026.
totan-nsc2026
Bone-age assessment and growth tracking for Thai children, built for NSC 2026 — medical imaging to prediction to an interface a clinician can actually use.
Different sensors, same problem: pull structure out of a noisy measurement of a human body. EEG is the next one.
Record
Loadout
Trajectory
Foundation — ML, Python, PyTorch
Linear and logistic regression through neural networks and gradient-boosted trees. The point isn't the models — it's understanding optimisation well enough that backpropagation stops being a black box.
Deep learning core
Backpropagation, CNNs, RNNs, and the practical side of training. Everything downstream — neural data analysis, modern RL — sits on top of this.
NeuroAI — RL and computational neuroscience
Bandits through policy gradients alongside neural coding and spiking models. Reinforcement learning is where the two halves of my interest turn out to be the same subject.
Transmit
Looking for a first research position in EEG / biosignal decoding
Undergraduate, unpaid is fine — I'm after the reps, not the stipend. Salaya, Bangkok, or remote. I've already built systems that pull structure out of RF, audio, and medical images; brain signals are the modality I actually want. If you run a lab working on neural data or anything adjacent, I'd be glad to hear from you.