Building ARISE
Luthira Abeykoon
I like making cool stuff, sometimes it’s electronics, sometimes it’s software, usually a mix of both. Into blending tech with creativity and seeing where it goes. Currently studying Electrical Engineering at the University of Toronto.
Work
Snapdragon display firmware team.
Building the all in one robotics education platform from CAD to code to augmented reality. Backed by North Forge, LAUNCH, and angels.
Data Foundation team,maintaining scalable data infrastructure, pipelines, and workflows for big data processing.
Developed forecasting models for solar and wind energy generation using Transformer architecture, and built API-driven data pipelines to support a deep reinforcement learning agent.
Pioneer workstudy SDL training labs team for Global Materials Discovery Lab team, building smart data pipelines to automate temperature control in next-gen material experiments
Core ERP software infrastructure team, autonomous OCR reciept processing and inventory management.
Simulations and data analytics for single-fiber action potential nerve propagation through non-invasive saphenous nerve stimulation for overactive bladder treatment.
Built OCR data pipelines automating manual data entry for 100,000+ land registry pins. (LRO #66)
Projects
Talos - Tensor Accelerated Logic for On-Chip Systems
Hardware accelerator for convolutional neural networks that runs both training and inference on chip.
Hackerfab
Making chips and transistors from the ground up.
CVChess
End-to-end model for chess state to FEN, 4× more accurate than SOTA.
AutoGrid AI
Microgrid management with deep reinforcement learning
DeepMetal
Compiler for high-level ML libraries to run your models on edge.
Zephron
RISC V Robot Arm with AXI communication and VGA dashboard.
A.I.M.
Artificially Intelligent Maps (w/ LLM powered Agentic Assistant)
Neural Signal Propagation Simulations
Single fiber action potential in the saphenous nerve.