Soumyajit De
Machine Learning Engineer at Meta
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolSoumyajit De is a machine learning engineer with 13 years of experience building scalable ML and data systems across major cloud and ads platforms, currently at Meta after a senior ML stint at Microsoft. He has driven measurable business impact—introducing click-prediction models and a global feature store that scaled ranking to 100+ markets and delivered multi-percent revenue and AUC lifts. His background spans production engineering at Oracle and performance-focused research at UCL, where he authored cache-friendly, multi-threaded algorithms that accelerated MMD-based tests by orders of magnitude. An active open-source contributor, he improved Shogun’s sparse-matrix log-determinant routines and implemented kernel-selection frameworks for MMD, reflecting deep numerical and kernel-method expertise not obvious from product roles alone. Combining low-level algorithmic rigor with large-scale deployment experience, he excels at turning research-grade methods into production ML services.
13 years of coding experience
13 years of employment as a software developer
B.Tech Computer Science and Engineering, B.Tech Computer Science and Engineering at Kalyani Government Engineering CollegeIndian Institute of Technology Bombay
Bengali, Hindi, English
Stackoverflow
315reputation
70kreached
4answers
2questions
Github Skills (19)
c-language10
numerical10
computation10
machine-learning10
numerics10
cprogramming-language10
numeric10
linear-algebra10
kernel-methods10
testing9
shogun6
depth-first-search6
php6
supervised-learning6
graph6
Programming languages (4)
Github contributions (5)
shogun-toolbox/shogun
Mar 2013 - Feb 2020
Shōgun
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:447 commits, 68 PRs, 152 pushes in 6 years 11 months
Contributions summary:Soumyajit's commits focused on adding and implementing a log determinant method for sparse matrices within the Shogun machine learning toolbox, suggesting a focus on improving the performance of Gaussian density likelihood estimation. This work involved adding new methods for computing the determinant of matrices within the mathematics module. Furthermore, the user has developed a framework for performing kernel selection for the MMD test.
cmakedata-sciencegunc-plus-plusmachine-learning
lambday/shogun
Jan 2015 - Feb 2020
The Shogun Machine Learning Toolbox (Source Code)
Contributions:2 PRs, 105 pushes, 14 branches in 5 years 1 month
toolboxmachine-learning-toolboxshogunmachine-learning
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.