Aswin Subramanian
Sr Applied Scientist at Microsoft
Redmond, Washington, United States
Summary
👤
Senior🎓
Top SchoolAswin Subramanian is a Senior Applied Scientist at Microsoft with eight years of experience building robust end-to-end speech systems, particularly for far-field, multi-talker environments. He holds a PhD from Johns Hopkins’ Center for Language and Speech Processing and a track record of research-to-product work across industry labs including MERL and internships at Tencent and NTT. At Microsoft he contributes to AI Cognitive Services’ E2E Speech Modeling team, and his open-source contributions to the widely used ESPnet toolkit added practical dereverberation (WPE, Beamformit) and improved realistic data preparation and evaluation pipelines. His background spans both research and systems engineering—from network software at Intel to applied ML research—giving him a rare ability to ship reproducible algorithms into production. Colleagues value his mix of rigorous academic grounding and pragmatic engineering that closes the loop between publications and deployable speech solutions.
8 years of coding experience
2 years of employment as a software developer
Indian Institute of Technology MadrasJohns Hopkins UniversityBachelor of Technology (BTech), Information Technology, Bachelor of Technology (BTech), Information Technology at Sri Sivasubramaniya Nadar College Of Engineering
Tamil, English
Github Skills (10)
pytorch10
speech-enhancement10
speech-recognition10
deep-learning10
python10
bash9
kaldi8
machine-translation7
speech-synthesis6
text-to-speech4
Programming languages (3)
Github contributions (5)
espnet/espnet
Jul 2018 - Nov 2018
End-to-End Speech Processing Toolkit
Role in this project:
ML Engineer
ML EngineerContributions:1 review, 19 commits, 2 PRs in 3 months
Contributions summary:Aswin primarily contributed to the end-to-end speech processing toolkit, focusing on the "reverb" recipe. Their work involved adding and integrating WPE (Weighted Prediction Error) and Beamformit algorithms, enhancing the speech enhancement capabilities. The user made modifications to data preparation scripts, including those for real and simulated data, to accommodate the new dereverberation methods. Additionally, the user added scripts to run WPE and Beamformit and updated the scoring and evaluation scripts.
speech-recognitionspeech-separationchainerspoken-language-understandingspeech-processing
sas91/academic-kickstart
Oct 2018 - Jan 2024
Contributions:46 pushes in 5 years 3 months
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.