Topic: Kernel optimization

Supports research on agentic, self-improving systems that generate and optimize kernels and end-to-end workloads on AWS Trainium. Topics include generative AI for kernel optimization, performance debugging and profiling, correctness and integrity evaluation, and supervised or reinforcement learning of open-source models for kernel and system optimization. Proposed research should connect generation, evaluation and learning loops and explain plans for open-source contributions. Use the ARA proposal template; four pages excluding appendices are encouraged. Decisions are expected in February 2027

Deep neural networks (DNNs) with the flexibility to learn good top-layer representations have eclipsed shallow kernel methods without that flexibility. Here, we take inspiration from deep neural networks to develop a new family of deep kernel method. In a deep kernel method, there is a kernel at every layer, and the kernels are jointly optimized to improve performance (with strong regularisation). We establish the representational power of deep kernel methods, by showing that they perform exact inference in an infinitely wide Bayesian neural network or deep Gaussian process. Next, we conje