BHUMI GODIWALA
Los Angeles, CA 984-***-**** ad167o@r.postjobfree.com https://www.linkedin.com/in/bhumigodiwala EDUCATION
University of Southern California Los Angeles, CA
Master of Science August 2021-May 2023
• Master of Science in Electrical and Computer Engineering (Machine Learning and Data Science specialization) (GPA 3.8/4.0) Dwarkadas J. Sanghvi College of Engineering, University of Mumbai Mumbai, India Bachelor of Engineering July 2016-October 2020
• Bachelor of Engineering in Electronics and Telecommunications Engineering (CGPA 9.19/10) TECHNICAL SKILLS
• Programming Languages: Python, Java, Git, C, C++, MySQL, SQL, Oracle, Object Oriented Programming (OOP/OOPs)
• Software: PyTorch, Jupyter Notebooks, Anaconda, PyCharm, Eclipse, Docker, JIRA, Node JS, AWS S3, AWS CLI
• Libraries and Frameworks: Matplotlib, Tensorflow, Keras, Numpy, OpenCV, Scikit-Learn, Pandas, Onnx
• Web-Technologies: HTML, CSS, Javascript, JSON
WORK EXPERIENCE
USC Information Sciences Institute Marina Del Rey, CA Machine Learning Researcher July 2023-Present
• Create ML models, researching distributed edge computing AI system design
• Investigate model compression, applying quantization, and pruning models for edge device enhancement TetraMem Inc Fremont, CA
Machine Learning Intern January 2023-May 2023
• Created ML models (e.g., Visual Wake Words, Human Pose Estimation) with PyTorch, Onnx Runtime. Verified on a state-of-the-art AI inference chip with in-memory computing capabilities
• Designed customized model with positive-only weights using PyTorch, maintaining accuracy. Applied Quantization Aware Training (QAT) for optimized models with reduced parameters
TetraMem Inc Fremont, CA
Machine Learning Intern May 2022-August 2022
• Built Human Pose Estimation ML models using PyCharm, Tensorflow, and COCO Python API in Linux environment. Validated on an in- memory computing AI inference chip post neural network optimization
• Executed post-training quantization to optimize developed Machine Learning models, achieving 93% accuracy with reduced size Tata Consultancy Services - ION Mumbai, India
Software Developer October 2020-August 2021
• Created JAVA, HTML, CSS, and JavaScript forms for university portals, deploying in real-time using TCS' framework
• Conducted metadata mapping, testing, reports generation, and optimized data segregation by course details ACADEMIC PROJECTS
ASL Gestures Prediction using ST-GAN for Shadow Removal
• Engineered pre-trained GAN-CNN fusion to enhance ASL Gesture classification, mitigating shadow effects
• Attained 92.9% test accuracy for 'E' and 'S' ASL Gestures classification. Utilized MLFlow for MLOps tracking and experience logging Banking Subscription Analysis
• Executed client subscription prediction using supervised algorithms: Logistic Regression, Decision Trees, Random Forests, and SVM
• Assessed performance via Accuracy, F1 score, and confusion metrics. Evaluated semi-supervised techniques including S3VM, label propagation, label spreading, and Co-training classifier Community Car Rental Platform
• Created car rental web app, deployed on Google Cloud CLI, and enhanced with Google Maps, Cloudinary, and VIN API integrations
• Designed database schemas via GraphQL with data hosted on MongoDB Movie Recommendation System
• Developed Movie Recommendation System: content-based, collaborative, and hybrid filtering. Preprocessed data using word embeddings and vectorization techniques
• Conducted EDA on MovieLens dataset for feature extraction and model evaluation using RMSE and MAE metrics Algerian Forest Fires Classification
• Spearheaded the creation of an advanced forest fire prediction framework employing a diverse array of classifiers including Bernoulli, SVMs, Decision Trees, and Random Forest. Diligently Evaluated performance through metrics including accuracy, F1 score, and confusion matrices
• Employed statistics, regression analysis, Principal Component Analysis (PCA), and Linear Discriminant Analysis (LDA) for insights Detecting COVID-19 with Chest X-Ray using PyTorch
• Classified Chest X-Ray dataset: Normal, Viral Pneumonia, COVID-19 using PyTorch ResNet-18 model
• Achieved an impressive 96% accuracy; conducted comprehensive analysis with Matplotlib, Torchvision, NumPy, PIL, and Python