Hello, I'm Sumit Kothari

Software Engineer & Data Scientist

Passionate about machine learning, data science, and building innovative software solutions. Graduated with my MS in Computer Science with expertise in predictive modeling and full-stack development.

View My Work Get In Touch
Full-Stack Development
Machine Learning
Data Science

About Me

Sumit Kothari

I am a dedicated Computer Science graduate student at the University of Southern California, with a strong foundation in software engineering and data science. My academic journey began with a Bachelor of Engineering degree in Computer Engineering from Mumbai University.

My passion lies in both software development and leveraging machine learning and data analytics to solve real-world problems. As a dedicated software developer, I have experience in developing predictive models, full-stack web applications, and conducting research in areas such as credit risk assessment and demographic prediction using computer vision.

MS Computer Science - USC
IEEE Published Researcher
AI/ML Specialist
1+
Years Experience
3+
Projects Completed
10+
Technologies Mastered

Technical Skills

Programming Languages

Python Java JavaScript TypeScript C++ SQL

Web Technologies

React.js Angular.js Node.js Django HTML5 CSS3 RESTful APIs

Machine Learning & AI

TensorFlow PyTorch Scikit-learn OpenCV YOLO NLP Computer Vision LangChain LangGraph

Databases & Cloud

MongoDB PostgreSQL MySQL Azure Git Docker

Education

Master of Science in Computer Science

Aug 2023 - May 2025

University of Southern California

Relevant Courses:
Analysis of Algorithms Database Systems Natural Language Processing

Bachelor of Engineering in Computer Engineering

Aug 2019 - May 2023

Mumbai University

Relevant Courses:
Java Data Structures and Algorithms Software Engineering Operating System Machine Learning Python

Professional Experience

Thrizer

Jan 2025 - May 2025

Software Engineering Intern

Los Angeles, California
  • Collaborated with a team of 5 developers to build and maintain Admin, Client, and Clinician portals, working on both frontend (Angular.js) and backend (Node.js, LoopBack) while following Agile methodologies on Azure DevOps
  • Developed and enhanced RESTful APIs using Node.js and LoopBack, integrating with MongoDB to meet client requirements
  • Managed and resolved user stories on Azure Boards, ensuring timely completion of sprint deliverables while using Git for version control
Angular.js Node.js LoopBack MongoDB Azure DevOps Git

USC Alzheimer's Therapeutic Research Institute (ATRI)

Mar 2024 - May 2025

Software Developer

Los Angeles, California
  • Developed 3 end-to-end projects, building 40+ RESTful APIs using Django REST Framework (DRF) while implementing React.js for the frontend and optimizing PostgreSQL database schemas
  • Implemented Test Driven Development (TDD) and Agile methodologies, employing Cypress for frontend testing, Django's built-in testing framework for backend testing, and Jira for task management to achieve high reliability and performance
  • Created a CI/CD pipeline using GitHub Actions to automate frontend, backend, and functional tests before deployment, enhancing the overall development workflow and deployment efficiency
Django React.js PostgreSQL Cypress GitHub Actions Jira

FPT Software

May 2024 - Aug 2024

AI Engineer Intern

Remote
  • Developed a Text2SQL agent to generate and execute SQL queries on custom databases using the GPT-3.5-turbo model and Azure OpenAI embeddings, incorporating client-specific information and business rules for improved quality
  • Utilized Azure Search to create and store embeddings, implementing semantic search for retrieval-augmented generation (RAG)
  • Led backend development and managed the team codebase using Azure DevOps for efficient collaboration and version control, and successfully deployed the bot on Azure cloud, integrating it with MS Teams using JavaScript for seamless user interaction
GPT-3.5 Azure OpenAI Azure Search JavaScript MS Teams Azure DevOps

VectorAB Labs

Jul 2021 - Dec 2021

Machine Learning Intern

Remote
  • Collaborated with a team of four to develop a parking lot entry system that detects cars and records license plate information, drawing insights from the analysis of more than 1,000 images of cars entering the parking lot
  • Implemented computer vision techniques: noise reduction, edge detection, and masking via OpenCV for improved accuracy in License plate detection
  • Executed vehicle detection and tracking through YOLO and DeepSORT, and integrated EasyOCR for license plate interpretation, resulting in improved accuracy and efficiency
OpenCV YOLO DeepSORT EasyOCR Computer Vision Python

Ignitus

Jun 2021 - Nov 2021

NLP Intern

Remote
  • Collected, cleansed, classified, and analyzed Reddit posts to predict individuals' depression status
  • Leveraged Pushshift.io API Wrapper (PSAW) to scrap 30,000 posts from 'depression' and 'SuicideWatch' subreddits, analyzing them with various Exploratory Data Analysis techniques like Tokenization, Lemmatization, Part-of-Speech tagging, Vectorization, etc
  • Compared and trained multiple machine learning models such as Support Vector Machine, Logistic Regression, Naïve Bayes, and ANN, achieving an accuracy rate of 84% through various optimizations and hyperparameter tuning
NLP Python Scikit-learn PSAW Machine Learning Data Analysis

Featured Projects

SkyQuery AI

Developed a UAVLogger Bot backend using FastAPI that enables users to query and analyze drone flight logs in natural language. Implemented a modular, multi-agent system using LangGraph, with specialized LLM-powered agents for each major telemetry key, allowing for targeted anomaly detection and flexible reasoning across UAV subsystems. Integrated MongoDB for efficient storage and retrieval of parsed MAVLink telemetry data, allowing scalable, low-latency access to flight records during LLM-driven analysis.

LangChain LangGraph MongoDB FastAPI Pymavlink OpenAI

SKBot - College Admin Bot

Designed a college admin bot using Python and PostgreSQL, providing comprehensive features for students, including attendance tracking, timetable access, and event management. Utilized the Python-Telegram-Bot API to create a responsive interface.

Python PostgreSQL Telegram Bot API PDF Generation

Credit Card Default Prediction

Machine learning model to predict credit card default risk using various algorithms including Random Forest, SVM, and Neural Networks. Achieved high accuracy in risk assessment.

Python Scikit-learn Pandas Machine Learning

Age, Gender & Ethnicity Prediction

Computer vision application using deep learning to predict demographic information from facial images. Implemented CNN architectures for multi-label classification.

Python TensorFlow OpenCV Deep Learning

ReadStop

Full-stack web application for book lovers to discover, track, and review books. Features user authentication, book recommendations, and social sharing capabilities.

React.js Node.js MongoDB Express.js

Publications & Research

Review Helpfulness Prediction using Deep Learning

Monali Shetty*, Prajakta Bhangale, Samarth Mehta, Sumit Kothari & Dr. Sharvari Tamane

Computer Integrated Manufacturing Systems, vol. 28, no. 11, pp. 673–685, November 2022

Research on predicting the helpfulness of product reviews using deep learning techniques, contributing to improved recommendation systems and user experience in e-commerce platforms.

Comparison of Age, Gender and Ethnicity Prediction Using Traditional CNN and Transfer Learning

Sumit Kothari, Samarth Mehta, Sujata Deshmukh

Published in IEEE Xplore, December 2022

Comparative analysis of traditional CNN architectures versus transfer learning approaches for demographic prediction from facial images, demonstrating improved accuracy in computer vision applications.

View on IEEE Xplore

Let's Connect

Get In Touch

I'm always interested in new opportunities and collaborations. Whether you have a project in mind or just want to connect, feel free to reach out!

LinkedIn GitHub sumit65kothari@gmail.com

Ready to work together?

Let's discuss how we can bring your ideas to life.

Start a Conversation