Frontier intelligence, security and edge systems — from agentic AI to post-quantum cryptography.
LLM agents that plan, act and self-correct
Encryption that survives quantum computers
Never trust, always verify
Train on patient data without moving it
Near-native, sandboxed compute everywhere
Treating images as sequences of patches
Brain-inspired, event-driven computation
Fine-tune billion-parameter models on one GPU
Hardening models against crafted attacks
Orchestrating ML models across the edge