Ben Datsko
PhD student in ECE at the University of Michigan. Computer architecture, physics-inspired computing, software–hardware co-design.
My group builds mixed-signal ICs that solve NP-complete problems — k-SAT, LDPC decoding — with relaxation oscillator arrays. The oscillators do the computation; a digital controller steers them. We get solutions in microseconds at a fraction of the energy of a CPU.
B.S.E. in CS from Michigan, 2025. Now pursuing the Ph.D. in ECE with Prof. Michael Flynn.
Recent work
MEDUSA A 200-variable k-SAT solver achieving 3.5× faster solution times and 3× greater energy efficiency than prior work. DACROQ Secure low-latency remote access platform for DARPA and NASA QuAIL, with hardware abstraction layer, API, and web frontend. DAEDALUS A continuous-time 3-SAT solver with unrestricted three-body spin interactions and 100% solvability on SATLIB benchmarks.