Gobinda Saha
AI Research Scientist at Meta
Menlo Park, California, United States
Summary
👤
Senior🎓
Top SchoolGobinda Saha is an AI research scientist with a PhD from Purdue and eight years of experience bridging deep learning algorithms and in-memory computing hardware. He designs neuro-inspired continual and lifelong learning methods—authoring innovations like Gradient Projection Memory, Scaled Gradient Projection, SPACE, and explainable episodic replay—that achieve near-zero forgetting and state-of-the-art gains on continual vision and RL benchmarks. Skilled in Python and PyTorch, he also brings practical hardware-algorithm co-design experience for electronic and photonic in-memory computing, including IMC primitives and HSPICE-anchored evaluation under nonidealities. Now at Meta in Menlo Park, he combines theoretical optimization (natural/second-order, bi-level methods) with hardware-aware training, model compression, and privacy-preserving decentralized learning. A less obvious strength is his consistent focus on explainability and resource-aware strategies that yield both accuracy and energy efficiency improvements in deployed inference scenarios.
8 years of coding experience
7 years of employment as a software developer
Doctor of Philosophy - PhD, Electrical and Computer Engineering, Doctor of Philosophy - PhD, Electrical and Computer Engineering at Purdue UniversityMaster of Science - MS, Electrical and Electronics Engineering, Master of Science - MS, Electrical and Electronics Engineering at Bangladesh University of Engineering and Technology
Bengali, English
Github Skills (4)
pytorch7
computer-vision7
deep-learning6
reinforcement-learning4
Programming languages (1)
Github contributions (5)
sahagobinda/CL_PCA
Jan 2020 - Nov 2021
Contributions:21 commits, 9 pushes in 1 year 9 months
sahagobinda/GPM
Mar 2021 - Jun 2021
Official [ICLR] Code Repository for "Gradient Projection Memory for Continual Learning"
Contributions:15 commits, 14 pushes, 1 branch in 3 months
continual-learningcomputer-visiondeep-learningoptimizaitonpytorch
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.