Vishwajitt Kamat

• Result-oriented professional with more than 2 years of extensive experience in Machine/Deep Learning, Data Engineering, NLP and Image processing.
• Integrated OpenAI and Anthropic large language models (LLMs) into core product features, enhancing the product's functionality and user experience.
• I am well-versed in utilizing computer vision techniques, such as image segmentation, object detection and reconstruction, SLAM and 3D Reconstruction among others with large datasets with different data types.
• Gained hands-on exposure to business operations, customer service, inventory management, and financial processes while assisting in day-to-day management activities.
• Developed skills in team coordination, problem-solving, and strategic decision-making, contributing to smoother operations and improved customer satisfaction.
• Designed and implemented API integrations for seamless communication between the company's platform and generative AI models.

• Date of Birth: 12/21/99
• Nationality: Indian

Insights Generation for survey

January 2023 – January 2025

This is the Product we as a company developed which provides a end to end solution for the survey industry. From survey initiation to insights generation of the survey is provided in the product. I as an engineer used LLM models to integrate it in the product for various NLP tasks such as insight generation, document parsing etc.

• Intergating LLM models from OpenAI and Antropic with the NLP tasks needed in the product.
• Use of Fast API for front-end data retrieval and Client user experience process.
• Use of Database Management and SQL language.
• Creating the Flow of the project and making a pipeline and package a end product.
• Integrating External API's with the product and optimizing the cost and frequency of api calls.

Document to survey programmable format

January 2023 – January 2025

• Developed an innovative solution to automate the conversion of document files (e.g., Word) into survey-programmable content for platforms like Confirmit and Decipher.
• Designed a robust pipeline using Python, natural language processing (NLP), regex and prompting for LLM and intgerating OpenAI and Antropic API through Langchain to parse and extract structured data from unstructured document formats.
• Implemented logic to convert extracted data into survey-ready formats, including JSON and XML, ensuring compatibility with survey tools.
• Integrated user-friendly features for handling complex survey elements, such as question branching, matrix grids, and validations.
• Optimized processing performance to reduce conversion time by 40%, by using async invoking in langchain improving productivity and streamlining survey programming workflows.
• Collaborated with stakeholders to refine tool functionality, resulting in a 30% reduction in manual survey creation errors.

Edge Enhancement Conditional GAN for Image Restoration

August 2021- December 2021

In this work, I proposed a deep neural network towards image restoration with high frequency details as a clue. Digital image restoration aims to recover an estimate of the original image from the degraded observations. The application of this project is in the fields of Underwater Imaging, Medical Imaging, Astronomy etc.

• Creating of a novel deep learning model.
• Mathematical formulation of the filters, Research and Development experience.
• Working with different state of the art models and Hands On on PyTorch.

SLAM Data Structure and Texturing core algorithm Development

january 2022- March 2022

Developed algorithm for structuring the data feed provided by edge device that involves RGB, Depth, IMU. Device exposed to involve iPhone LiDAR, Intel Real sense and proprietary in house device. I developed Preprocessing programs for both depth and RGB data for proprietary device with data type conversion for SLAM required formats. Here we also used Camera Trajectory generation and point cloud optimizations within SLAM algorithm thresholds and developed a Methodology for fast and effective texturing for SLAM based 3D reconstruction.

• Handling 3D point cloud data, building and refining 3D applications.
• Competence with computer vision libraries like Open3D, OpenMVG, OpenMVS.

Other Major Development Involves

January 2022- April 2023

• Camera pose Estimation in Multi-view Environment
• Video to 3D model with Deep Learning and Photogrammetry
• Path/Trajectory construction and Loop Closure Analyses with SLAM-based optimization algorithms.
• Point-Cloud creation using stereo pair of cameras with camera intrinsic and extrinsic validation.

• Table Tennis
• Badminton
• Cricket

• English
• Hindi
• Kannada
• Konkani