Vineel Pratap
Member Of Technical Staff at WaveForms AI
Menlo Park, California, United States
Summary
🤩
Rockstar🎓
Top SchoolVineel Pratap is a research-oriented software engineer with 12 years of experience building scalable machine learning systems, now serving as a Member of Technical Staff in Menlo Park. He spent nearly a decade at Facebook as a Research Engineer working on speech and multilingual ASR within fairseq’s MMS project, improving alignment, documentation, and usability while integrating large‑scale language models. His open-source contributions include CUDA-focused enhancements to the C++ flashlight ML library and practical tooling for fairseq, reflecting deep familiarity with both low-level performance engineering and applied ML research. Vineel holds an MS in Computer Science from UC San Diego and a BTech from IIT Bombay, combining strong academic foundations with production engineering. Colleagues rely on him to bridge research prototypes and production-quality distributed training on CUDA hardware. He brings a pragmatic focus on developer experience—tutorials, Colab notebooks, and docs are a recurring part of his impact.
12 years of coding experience
9 years of employment as a software developer
University of California, San DiegoIndian Institute of Technology Bombay
Github Skills (17)
pytorch10
c-language10
python10
machine-learning10
voice-recognition10
automatic-speech-recognition10
asr10
cuda10
speech-recognition10
nlp10
cprogramming-language10
fairseq10
notebook9
neural-network9
deep-learning9
Programming languages (6)
Github contributions (5)
flashlight/flashlight
Dec 2018 - May 2021
A C++ standalone library for machine learning
Role in this project:
Back-end Developer
Back-end DeveloperContributions:1 release, 2 reviews, 64 commits in 2 years 5 months
Contributions summary:Vineel primarily contributed to the `flashlight/flashlight` repository by modifying the `flashlight/distributed/backend/cuda/DistributedBackend.cpp` and `flashlight/nn/modules/Activations.h` files. These changes include adding CUDA-specific functions and modifying activation functions, which aligns with the repository's description of being a machine learning library. The changes suggest a focus on improving the library's functionality and efficiency for distributed machine learning tasks, particularly on CUDA-enabled hardware.
cppdeep-learningstandalone-libraryc-plus-plusmachine-learning
facebookresearch/fairseq
Mar 2022 - Mar 2022
Facebook AI Research Sequence-to-Sequence Toolkit written in Python.
Role in this project:ML Engineer
ML EngineerContributions:14 reviews, 1 commit, 33 PRs in 1 day
Contributions summary:Vineel contributed to the MMS (Massive Multilingual Speech) project within the fairseq repository, focusing on speech-related tasks. Their contributions included fixing issues in MMS alignment code, adding a tutorial for language identification (LID) inference with a Colab notebook, and integrating a Common Crawl language model for improved ASR accuracy. Furthermore, the user made various updates to documentation and tutorials related to the MMS ASR model, demonstrating a focus on usability and accessibility.
pytorchnlpsequencepythontransformer-architecture
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.