Electronics · Embodied Intelligence · Neuro
Currently an undegraduate student at the University of Toronto studying Engineering Science, majoring in ECE and minoring in Robotics.
I'm building intelligent systems at the intersection of AI, neuroscience, and robotics. I design embedded hardware for brain–computer interfaces, develop AI agents for physical control in BCI applications, and study how learning emerges from constraint.
Right now, I’m focused on embodied intelligence and neural interfaces, training reinforcement learning and imitation learning agents to control real systems, while designing hardware that senses, responds, and adapts to brain-derived signals. I’m particularly drawn to low-latency, neuromorphic platforms where inference happens on-device, in real time.
I care about creating systems that are both functional and elegant: from wearable neural loggers and EVSE boards, to Transformer-based control policies and world models for motor drivers. I see engineering as an artistic process, where constraints force creativity and the medium is physical.
Some things I'm interested in: cultural identity and formation, intelligence, intuition, and artistic expression. I'm excited by ideas that let you build for the body and the mind with tools that feel as intuitive as movement, deliberate as thought, and expressive as art.
In the long term, I plan to center my work around early developmental learning (when the brain is the most malleable). I also hope to contribute to the development of solutions for neurological disorders that surface in childhood. I’m drawn to the subtle art of shaping cognition and not by force, through understanding how growth unfolds when gently guided.
ConSENS Lab — Developing evolutionary algorithms to infer L2/3 microcircuit dynamics from EEG signals, under PhD Romesa Khan, 2025–
Intelligent Microsystems Lab — RF board design for wireless neural implants, under PhD Mohammad Abdolrazzaghi, 2025–
UofT Aerospace Team — Software lead, 2024-2025
LUNR Corp — Flight systems intern, neural fin generation & GCS dev, 2024
Major League Hacking — Top 50 Hacker, 2023
Our Wave Hub — Fullstack developer intern, 2022
HackTheNorth — Finalist, 2022
Beauty of Mathematics — Author, 2021
World Model-Based Reinforcement Learning for Real-Time Motor Control
World model RL agent trained in PyBullet and deployed to STM32 for torque control with < 0.2 ms inference.
[Paper]
[Board]
[Code]
Transformer-Based Edge Agent for Real-Time Fault Detection in Smart Grid Nodes
Self-supervised Transformer deployed to embedded node for real-time power grid anomaly detection (AUC > 0.90).
[Paper]
[Board]
[Code]
Reinforcement Learning for Embedded EV Charging Control: Sim-to-Real Optimization of Relay Policies via PPO
PPO agent trained to optimize EV charging control under relay timing, thermal, and fault constraints.
[Paper]
[Board]
[Code]
Spatially-Informed Neural Compression via GCN-Enhanced Transformers for BCI Systems
EEG + IMU → Transformer → Action command. Visualizes attention maps and exports to ONNX.
[Paper]
[Code]
Sim2Real Transformer for Proprioceptive Contact Estimation
Transformer trained on simulated IMU + encoder data to infer contact states across robot morphologies, with MMD adaptation and STM32 deployment.
[Paper]
[Board]
[Code]
Tesla BLE EVSE Board
STM32-based EV charger with BLE telemetry, pilot sensing, relay control, and thermal/GCP fault logic.
[Technical Writeup]
Smart Grid Node with Remote Fault Detection
Precision voltage/current sensing with BLE telemetry, RTC-GPS sync, and transformer-based fault detection.
[Technical Writeup]
Tesla Roadster Infotainment System
Redesigned 8-layer board for media/audio interface, CAN I/O, and UX improvements over open-source layout.
[Technical Writeup]
NeuroStimCore
Closed-loop stimulation system for neuroadaptive feedback with STM32 control and fault-protected outputs.
[Technical Writeup]
CDAQ-EEG
Compact STM32-based EEG acquisition board with BLE streaming, microSD logging, and CNN-based neural decoding.
[Technical Writeup]
ASICFakeout
6-layer test board for ASIC emulation and FPGA validation, with multi-rail power, clocking, and SPI hooks.
[Technical Writeup]
SmartInterfaceCore
Embedded interface bridge with CAN isolation, microSD logging, trip logic, and STM32-based sensor routing.
[Technical Writeup]