Rishi Kesh
Co-Founder And CTO at dubpro.ai
Delhi, India
Summary
👤
Senior🎓
Top SchoolRishi Kesh is a Co-Founder and CTO with 11 years of hands-on experience building production-grade AI systems, specializing in agentic and generative models, LLM alignment, and speech synthesis. Based in Delhi, he leads dubpro.ai’s efforts to create human-like text-to-speech and voice transformation models, bringing research-grade techniques into scalable products. His open-source contributions include implementing HiFi-GAN components in the widely used coqui-ai TTS toolkit and improving scikit-learn examples and docs, reflecting both deep model expertise and attention to developer usability. Earlier roles span end-to-end speech engines, reinforcement learning for dialogue, knowledge graph construction, and high-performance product-matching systems, showcasing a rare blend of research, ML engineering, and full-stack software experience. Collected technical breadth—from Wavenet and DeepVoice to deployment automation—enables him to move complex ML innovations from prototype to production.
11 years of coding experience
4 years of employment as a software developer
Kendriya Vidyalaya SangathanBachelor of Technology - BTech Electrical Electronics and Communications Engineering, Bachelor of Technology - BTech Electrical Electronics and Communications Engineering at National Institute of Technology Silchar
English, Hindi
Stackoverflow
71reputation
3kreached
3answers
2questions
Github Skills (33)
pytorch10
python10
matplotlib10
scikit10
machine-learning10
text-to-speech10
generative-adversarial-network10
ml10
mle10
gtts10
freetts10
deep-learning10
trainings10
scikit-learn10
modeling10
Programming languages (10)
Github contributions (5)
scikit-learn/scikit-learn
Feb 2017 - Jun 2017
scikit-learn: machine learning in Python
Role in this project:
Data Scientist
Data ScientistContributions:10 commits, 14 PRs, 56 comments in 4 months
Contributions summary:Rishi primarily contributed to improving the examples and documentation within the scikit-learn repository. Their work included updating examples to be compatible with newer versions of Matplotlib, specifically by adding the `edgecolors` attribute to scatter plots. Additionally, the user fixed formulas in the Gaussian Process kernel documentation, improved visualization in example plots, and enabled code coverage reporting using Codecov. This indicates a focus on enhancing the usability, visual clarity, and maintainability of the machine learning examples.
machine-learningpythonscikit-learnstatisticsdata-science
coqui-ai/TTS
Feb 2021 - Mar 2021
🐸💬 - a deep learning toolkit for Text-to-Speech, battle-tested in research and production
Role in this project:ML Engineer
ML EngineerContributions:8 commits, 1 PR, 4 comments in 19 days
Contributions summary:Rishi primarily contributed to the development of the HiFi-GAN vocoder within the TTS project. Their work involved implementing HiFi-GAN generator and discriminator classes, integrating MSD with the Multi-Period Discriminator, and adding Exponential LR for training. They also addressed minor bugs and made code trainable, suggesting a focus on model architecture, training, and optimization within the text-to-speech framework.
deep-learningtext-to-speechpythonspeechpytorch
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.