Jeffrey Seelan

• Designed and developed an Large Action Models and Multi Agent Framework , Central Tool layer as a top interface layer with RAG for all integrations inside Zoho which improved accuracy 2x when compared to the zero shot LLM calls .
• Integrated Pretrained Visual Foundational Models into zoho products . Added support for fine tuning , which improved the accuracy 20% than the traditional supervised approaches against each cv task.
• Integrated Video Foundational Models for optimizing the search across zoho understanding the visual context .
• Working on Retroformer for Agent framework and Agents - Large Visual Models integration

.

• Built Retail product's category prediction based for ONDC (open network for digital commerce) for migrating all our current Point of Sales - item schema in Gofrugal to central ONDC data schema based on NLU , ResNet, YOLO .
• Built Churn prediction and Customer bucketing solution to Gofrugal's CRM for targeted campaigning , and Customer Experience management closely working with Support teams.
• Researched and implemented Beta Amazon-Go like store internally inside Gofrugal for employees to buy stuffs using Pose Detection , Object Detection , Stereo scenic vision, 3d reconstruction .

• Build hybrid recommender system with ResNet/TF-IDF/WordVec based recommendation and Collaborative Filtering based approaches for Gofrugal's ecommerce solution.
• Built Lead Scoring for Gofrugal CRM's solution with Clustering / distance based approaches .
• Deployed ML systems in Flask , FastAPI frameworks in Docker and Docker compose environments hosted in AWS EC2s.

• Built a social media post sensitivity analyzer based on the website session's data
• Constructed ETL pipeline where extracted data from various social media sources with web scraping tools .
• Created visualizations of overall social trend with PowerBI