Kaushal Trivedi
Co-Founder And Chief Technologist at utterworks
England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolKaushal Trivedi is a Co-Founder and Chief Technologist with eight years focused on AI/ML productization and a 20+ year background in enterprise CRM and SAP solutions. He has built and shipped production deep learning platforms, a cross-channel conversational UI that automates 80% of high-volume transactions for a major UK energy firm, and an AI-powered cognitive search and call QA platform used in enterprise settings. Kaushal is the creator of the popular open-source Fast-BERT library and holds a patent for a Natural Language Interface for data management, reflecting his ability to turn cutting-edge research into practical products. He combines hands-on ML engineering—integrating transformers and LLMs like OpenAI, Bedrock and Llama—with strategic leadership across presales, multi-site delivery and cloud architecture. His work has driven over a million pounds in company revenue and earned industry awards for conversational AI, demonstrating both commercial and technical impact. Fluent at bridging business stakeholders, cloud vendors and engineering teams, he excels at shipping enterprise-grade cognitive automation.
8 years of coding experience
7 years of employment as a software developer
Bachelor of Engineering, Information Technology, Bachelor of Engineering, Information Technology at University of Mumbai
Github Skills (9)
transformers10
pytorch10
fastai10
machine-learning10
nlp10
python10
machine-learning-models10
bert10
multilabel-classification9
Programming languages (2)
Github contributions (5)
utterworks/fast-bert
Apr 2019 - Apr 2022
Super easy library for BERT based NLP models
Role in this project:
ML Engineer
ML EngineerContributions:5 releases, 110 commits, 61 PRs in 3 years
Contributions summary:Kaushal primarily contributed to the development of a library for BERT-based NLP models. They implemented core functionalities within the library, including a learner class, model definitions for multi-label classification, and a data bunch for handling BERT-specific data structures. Their work involved integrating fastai and pytorch-pretrained-bert, as well as sample notebooks for demonstrating usage, indicating a focus on making BERT models accessible and easy to use. The changes included model definition, the implementation of training routines, validation and metrics.
fast-bertnlptransformersdistilbertbert
kaushaltrivedi/bert-toxic-comments-multilabel
Feb 2019 - Aug 2019
Contributions:7 commits, 6 pushes, 1 branch in 6 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.