Dhruv Kumar
Member Of Technical Staff at Cantina Labs
Vancouver, British Columbia, Canada
Summary
👤
Senior🎓
Top SchoolDhruv Kumar is an applied research scientist and Member of Technical Staff with 11 years of experience building production-grade NLP systems and research-driven ML models. He led development of text generation and paraphrasing products at Grammarly, open-sourced the CoEdIT family of instruction-tuned LLMs (770M–11B) extended across six language families, and has published multiple ACL/EMNLP/NAACL papers plus patents. His background blends academic rigor from a University of Waterloo CS master’s with hands-on engineering across startups and finance, including contributions to tensorflow/datasets where he added NLP-focused datasets like wiki_auto and asset. Comfortable moving models from research to product, he’s also experienced in RAG, tool integration (calendars/KBs), and mentoring research interns on evaluation and personalization. Based in Vancouver, he combines deep model-building expertise with practical deployment experience and a track record of creating reusable open-source resources.
11 years of coding experience
7 years of employment as a software developer
Bachelor’s Degree Information Technology, Bachelor’s Degree Information Technology at Indian Institute Of Information Technology Allahabad ISC, ISC at St. Francis CollegeMaster's degree Computer Science, Master's degree Computer Science at University of Waterloo
English, Hindi
Github Skills (11)
machine-learning10
tensorflow10
python10
data-integration10
datasets10
data-set10
user-data9
data-model9
nlp9
jax3
numpy3
Programming languages (5)
Github contributions (5)
tensorflow/datasets
Dec 2020 - Jan 2021
TFDS is a collection of datasets ready to use with TensorFlow, Jax, ...
Role in this project:
ML Engineer
ML EngineerContributions:9 commits, 2 PRs, 4 comments in 28 days
Contributions summary:Dhruv's contributions primarily focus on adding and modifying datasets within the `tensorflow/datasets` repository. They introduced the `wiki_auto` and `asset` datasets, implementing the necessary code for data loading, preprocessing, and integration with the TensorFlow Datasets framework. These changes demonstrate a focus on enabling the use of simplification datasets for training and evaluating machine learning models, particularly in the domain of natural language processing.
datanumpydeep-learningdatasetmachine-learning
ddhruvkr/Consentio
May 2019 - Oct 2020
Contributions:9 commits, 18 pushes, 1 branch in 1 year 4 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.