About me

I'm a Machine Learning Engineer based in Los Angeles, CA, with expertise in computer vision, natural language processing, and large language models. Proficient in Python, TensorFlow, and OpenCV, I bring a strong technical foundation and a passion for innovation to every project.

My research has focused on fine-tuning large language models for real-time applications and enhancing user engagement through AI-driven mental health chatbots. By optimizing human-AI interaction and system efficiency, I strive to create impactful solutions that merge cutting-edge technology with real-world needs.

As a lifelong learner and advocate for diversity in tech, I'm committed to fostering an inclusive environment where creativity thrives and innovation knows no bounds. Let's connect and build something amazing together!

Resume

Education

  1. University of Southern California, Los Angeles

    2022 — 2024

    Master in Science, Compter Science

  2. University of Pune, India

    2018 — 2022

    Bachelor of Engineering, Computer Engineering

Experience

  1. Machine Learning Researcher

    University of Southern California

    January 2024 — Present
    • Leveraging Mistral-7B and Google's Gemini for fine-tuning on Hugging Face, reducing model response latency during navigation command inference by 20%.
    • Developing mood detection and context-aware systems using transfer learning to enhance driver-vehicle communication.
    • Designing cross-platform in-car interfaces with Flutter APIs, integrating LLMs for real-time navigation and diagnostics.

    Tech Stack: Python, Machine Learning, Large Language Models (LLMs), Natural Language Processing (NLP)

  2. Learning Assistant

    University Of Southern California

    August 2023 — May 2024
    • Courses: Programming in Python, Accelerated Python
    • Tutored class of 40 students with non-programming backgrounds by offering personalized guidance resulting in 10% better understanding of practical applications.
    • Created weekly assignments based on board games to help students understand and grasp concepts better.

    Tech Stack: Python

  3. Machine Learning Engineer Intern

    Vlinder, Inc

    June 2023 — August 2023
    • Researched OCR and image processing techniques, reducing counterfeit label fraud by 35% and improving product authentication.
    • Implemented histogram comparison and template matching algorithms, achieving 92% accuracy in text recognition.
    • Optimized deep learning pipelines for edge detection and statistical analysis, enhancing model efficiency and reliability.

    Tech Stack: Python, Machine Learning, Optical Character Recognition (OCR), Computer Vision (CV)

  4. Software Development Intern Lead

    Metalympics Limited

    June 2023 — August 2023
    • Architected relational database structure on Amazon RDS for “Fraxir” platform, ensuring 80% optimized data management.
    • Elevated website performance by performing Performance Profiling, achieving 60% increase in speed and responsiveness.
    • Mentored 3 junior developers using daily Scrum methodologies in the development and launch of Fraxir, online education platform focused on Web3 and AI technologies.

    Tech Stack: HTML, CSS, JavaScript, Node.js, Amazon Web Services (AWS)

  5. Software Engineer Intern

    Radon Tech

    August 2020 — November 2020
    • Deployed KNN-based recommendation engine, improving user experience by 25% and boosting sales by 20%.
    • Conducted A/B testing and debugging to refine Machine Learning pipelines, resolving 70% of deployment issues and enhancing system reliability and user satisfaction.
    • Collaborated with cross-functional teams to deliver “BeBig” e-commerce mobile app using Flutter, implementing CI/CD pipelines and Agile methodologies for efficient deployment.

    Tech Stack: Flutter

  6. Software Developer Intern

    Edify Accelerators

    April 2020 — July 2020
    • Orchestrated team of 4 for “Bloom India” Blood Donation Android app facilitating quick donor-receiver connections.
    • Designed and implemented chat and geolocation features for precise donor matching and communication based on proximity boosting growth by 40%.
    • Revamped app architecture to user-friendly interface improving navigability and user experience by 20%.

    Tech Stack: Android Development

Publications

  1. Mental Health Mobile Application with Diagnosis, Sentiment Analysis and Chatbot

    April 2022

  2. Sentiment Analysis using Chatbot and Mental Health Tracker

    December 2021

Certifications

  1. Generative AI with Large Language Models

    June 2024

  2. Machine Learning by Andrew Ng

    Sept 2020

Projects

  • Chatbot

    CarRentAI Chatbot

    • Designed an advanced Q&A chatbot for car rental services using Google PaLM, achieving 95% query response accuracy.
    • Delivered an intuitive UI with Streamlit utilizing LangChain for LLM management, boosting user engagement rates by 30%.
    • Employed FAISS vector database and Hugging Face embeddings to optimize data retrieval, reducing response time by 40%.

      • Tech Stack: Python, LangChain, Google PaLM, Hugging Face, FAISS

  • Tic-Tac-Toe

    Tic-Tac-Toe AI

    • Engineered a Tic Tac Toe AI leveraging Alpha-Beta Pruning for strategic move optimization.
    • Developed a Tic Tac Toe AI algorithm using Alpha-Beta Pruning for enhanced gameplay strategy.

      • Tech Stack: Python, Minimax

  • Shazam-Video

    Shazam-Video

    • Developed video identification system inspired by Shazam, leveraging OpenCV for video processing and Librosa for audio analysis, achieving 85% accuracy and 40% faster identification.
    • Implemented shot boundary detection, motion analysis using RGB, and audio analysis enhancing indexing and matching precision by 25%.
    • Integrated algorithms into unified system achieving 30% optimization in pattern recognition and processing efficiency.

      • Tech Stack: Python, Computer Vision, Video Processing

  • Frost

    Martian Frost

    • Conducted a project on the identification of frost in Martian HiRISE images using advanced image processing techniques.
    • Developed a deep learning image classification model using transfer learning with EfficientNetB0, VGG16, and ResNet50 architectures.
    • Compared traditional CNN and transfer learning models to evaluate performance differences on small datasets.

      • Tech Stack: Python, Computer Vision, Transfer Learning

  • Pente-AI

    Pente AI

    • Coded AI bot for human-versus-machine gameplay in Pente using Python.
    • Formulated alpha-beta pruning algorithm for AI bot with 87% success rate against humans.
    • Tested bot achieving 60% win rate and 80% faster game completion than traditional methods.

      • Tech Stack: Python, Minimax

  • HDR

    Handwritten Digits Recognition

    • Leveraged Pillow with TensorFlow to develop Neural Networks (CNNs) classifier, achieving 87% accuracy in handwritten digit recognition.
    • Optimized model performance through hyperparameter tuning, reducing training time by 40%.
    • Built Flask-based interactive web application with MySQL backend for real-time digit prediction and data storage.

      • Tech Stack: Python, FLask, Neural Networks, HTML, CSS, JavaScript, SQL