Summary
Marcelo Do Nascimento is a senior researcher at Microsoft focused on neural network compression and efficient AI for edge and real-time deployment. He earned a DPhil in Engineering Science at Oxford (Active Vision Lab, supervised by Prof. Victor Prisacariu) in 2022, after completing an MEng in Engineering Science; his graduate work centered on compressing CNNs to run faster with less memory on FPGAs in collaboration with Omnitek (now Intel). At Microsoft, his work spans distillation, pruning, and quantization for super-resolution networks at the edge, pruning of large language models, and compression of diffusion models for real-time edge image generation. Based in London with around a decade of experience, he effectively bridges academic research and production-scale AI systems. He has a track record of collaborating with industry sponsors and translating cutting-edge neural network compression techniques into practical, deployable solutions.
11 years of coding experience
2 years of employment as a software developer
DPhil, Egineering Science, DPhil, Egineering Science at University of Oxford