Summary
Subhankar Pal is an ML Research Scientist with nine years of experience at the intersection of algorithms, computer architecture, and VLSI, currently enabling hardware-aware CV and LLM models for Meta Reality Labs' AR initiatives. He holds a PhD from the University of Michigan and has led end-to-end efforts from RTL and microarchitecture design to compiler-driven dataflow optimizations for privacy-preserving computation such as homomorphic encryption. At IBM he prototyped an FHE accelerator frontend, devised PiM-aware scheduling, and developed SoC exploration tools that accelerate convergence for real-time workloads. His background spans GPU bring-up at NVIDIA to instruction-fetch improvements at AMD, giving him rare full-stack insight from silicon to ML models. Known for practical research tooling—like the HetSim simulator and contributions to STOMP—he blends rigorous academic rigor with production-driven engineering for next-generation ML hardware.
9 years of coding experience
10 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science and Engineering, 4.0/4.0, Doctor of Philosophy - PhD, Computer Science and Engineering, 4.0/4.0 at University of Michigan
Bachelor of Engineering - BE, Electrical and Electronics Engineering, 9.61/10, Bachelor of Engineering - BE, Electrical and Electronics Engineering, 9.61/10 at Birla Institute of Technology and Science
English, Bengali, German, Hindi, Telugu