1. Bittensor (Deep Learning Model Optimization) :Developed a strategy to construct a compact neural network from a larger architecture, ensuring comparable accuracy and lower latency. Achieved a remarkable reduction of over 96.67% in model parameters (from over 30 million to 1 million) with a minimal accuracy drop of approximately 9% for image classification tasks. Concurrently, I am leading a team of machine learning developers, assisting them with their assigned tasks and managing daily stand-ups to track progress to keep higher management informed.Additionally also contributing to various projects by supporting the team and executing designated tasks, demonstrating versatility and cooperation.

1. Dental Segmentation: Utilized deep neural net architecture (U-Net) to design and train a DL model for predicting key dental landmarks such as ABC, CEJ, Bone, and gingiva etc. This project aimed to enhance dental imaging analysis through deep learning techniques.Achieved outstanding precision with an average surface distance error under 20 microns and an average point distance error under 60 microns for 1800x800 dimension images.2. Slack Standup Bot: Engineered a Slack bot utilizing FastAPI, saving more than 80% of the time typically spent on writing daily updates, through automatic routing of standup updates to designated channels. Implemented automatic formatting, reminders, and centralized data storage for real-time management access. Additionally Integrated machine learning techniques to detect repetitive standups and notify developers, team leads, and management. Enabled instant insights generation of developer and project progress via LLM integration, streamlining the overall update process. 3. PDF to Excel Report Generator:Developed a robust tool using Streamlit, PyMuPDF, AWS Textract, and large language model to automate invoice-to-Excel conversion.The workflow involves raw file conversion, optional human intervention for edits, user approval, and final Excel report generation with predefined aggregation functions. This solution achieved an 85% reduction in the time needed to maunally produce reports, eliminating format dependency.Managed all of the above projects single-handedly along with client calls, ensuring seamless coordination and timely delivery. Additionally also contributed to feasibility research for diverse AI/ML inquiries.

Developed and implemented web scraping solutions using Python, leveraging tools like Selenium, BeautifulSoup, and APIs. Efficiently extracted datasets from various websites, ensuring data accuracy and compliance with web scraping guidelines.Additionally received extensive training in odoo.

Petpooja, a leading provider of POS solutions for restaurants, launched an Android application named 'tvito,' which features thousands of designs for various food cuisines.I contributed over 1,000 designs for the same.

Engaged as a Tele Marketer and Cold Caller for Hungrito Food Fest 4.0, reaching out to people to promote the event and boost attendance. Additionally, played a pivotal role in the on-ground operational team during the festival, contributing to the seamless execution of the largest food fest in Ahmedabad city.