Piyush Kumar

Powering next gen Acrobat features with On Device ML models

Worked as Software Engineer AI/ML at Adobe.

Worked as Lead Data Science Mentor at DevTown.

- Developed Deep Learning Models for Handwritten text recognition in banking sector
- Declared winner in the competition and awarded with a cash prize of INR 300k

- Developed ML models for American Express credit card default time prediction
- Declared 1st runner up in the competition which saw over 3600+ participation across 21 top institutes of India.

Worked with the AI and Machine Vision team on Document Digitization and Token Classification
• Built Document AI models for Form Recognition, Receipt and Invoice Understanding using LayoutLMv3 with pytorch and transformers and automated the process of extracting relevant data from scanned and pdf documents
• Trained the model on FUNSD and CORD datasets with 4 and 23 classes to achieve an F1 score

Worked as Editor at IIT Tech Ambit.

◦ Secured 4th place in the global intercollegiate finals among over 4000+ teams across 22 countries
◦ Provided key business solutions for The Walt Disney Company across 2 platforms: Disney Web and Disney App for driving up the average order value using Adobe Analytics Workspace
◦ Led a team of three and provided 4 solutions, namely Analysis, Redesign, Execution and Connect with an estimated growth

Worked as Research Intern at Indian School of Business.

• National Winner of ABInBev’s Maverick 2.0 Hack-a-thon among 750+ teams all across India. The only team of sophomores to make it to the final. Awarded with a cash prize of INR 1 lac.
• Built a recommendation system for cross-sell and upsell of ABInBev’s products. Used Neural Collaborative Filtering for building the recommendation system. LSTM model was used to recommend quantities for the same.
•

Worked as Data Science Intern at SirpiDataScience.

Issued Jan 2026 · Expires Jan 2029

Major in Instrumentation Engineering with Specialization in Artificial Intelligence

Issued Dec 2020