Hardware/GPU × LLM
FlashKernel
CUDA and Triton kernels for attention, activation fusion and cache operations. Explore the implementation, correctness boundaries and GPU validation requirements.
Flagship Projects
Four projects, with code and open questions
GPU inference, robotics, EEG and energy optimization: explore the implementations, available evidence and next validation steps.
LLM × Robotics × GPU
RoboLLM
A MuJoCo prototype tested on fixed-goal reaching and one-second holds, with collision checks, complete traces and reproducible controller comparisons.
BCI × LLM × GPU
NeuroLLM
An EEG transformer prototype for masked reconstruction and motor-imagery classification, with data separation and evaluation as the current priorities.
Quantum AI × Energy
QuantumGrid
Synthetic unit-commitment experiments using QUBO, QAOA, VQE and classical solvers. Explore the formulation and requirements for a fair comparison.