Name: Manavi Ghorpade
Profile: Software Engineer
Email: ghorpademanavi@gmail.com
Phone: 530-591-5243
Degree: M.S. in Computer Science
Skill
C++About me
Hey there! Welcome to my little corner of the internet. I'm a proud Computer Science graduate from California State University, Chico. I’ve got a solid grip on Data Structures and Algorithms, and I love tackling problems with a creative twist. Whether it's cracking a complex algorithm or solving a brain-busting puzzle, I live for that "Eureka!" moment.
I’m fluent in C++ & Python.
When I’m not lost in code:
Mountains are my playground: I love hiking and finding new trails.
Dance like nobody's watching: Seriously, I’ve got moves.
Furry friends: Volunteering at the local animal shelter is my happy place.
Publications
1. Diabetes Detection System
(Click here)
2. A Doctor's Appointment Booking System using Recommendation Model (Click here)
Work Experience
Research Assistant
California State University, Chico, CA | Jan 2024 - Present- Developing independent Large Language Model (LLM) and Small Language Model (SLM) like Llama-3 and Phi-3 for student interaction
- Implemented web scraping using Beautiful Soup to gather data and stored collected data in Chroma vector database
- Utilized data preprocessing and augmentation techniques to enhance the robustness of training datasets
- Optimized LLM performance through fine-tuning with QLoRA, and implemented guardrails for model accuracy and control over output
- Implemented RAG on Llama 3 and Phi 3 architectures and improved the performance by 30%
- Utilized cloud computing platforms like AWS and GCP for scalable model training and deployment
- Coordinated with faculty and student teams to align research objectives with university goals and student needs
- Provided mentorship to undergraduate students working on related projects, fostering a collaborative research environment
Student Assistant
California State University, Chico, CA | Aug 2022 - May 2023- Tutored 5-6 students weekly in computer science, focusing on data structures, algorithms in C++ programming language
- Coached students during 60+ hours each semester, fostering independence in problem-solving, coding, and debugging skills
- Provided real-time coding demonstrations and hands-on exercises to reinforce theoretical concepts
- Assisted students with the use of development tools, version control systems (Git), and integrated development environments (IDEs)
Information technology Intern
Priar Technoworld, Mumbai, India | May 2020 - Nov 2020- Collaborated closely with students, offering hands-on learning sessions in Linux fundamentals, Shell Scripting and AWS
- Assisted students in diagnosing and troubleshooting queries related to installation, configuration, and command-line environment
- Provided support for AWS cloud services, including EC2, ELB, AWS SageMaker, ASG, S3, ECS, VPC, Lambda, SQS, and SNS
- Developed and maintained comprehensive documentation and tutorials to aid students in their learning journey
- Organized and led study groups and coding bootcamps to prepare students for exams and certifications
- Assisted in the creation and deployment of sample projects to demonstrate real-world applications of course material
Projects
Video Game Sales Analysis
Technologies used: BeautifulSoup, Python, Tableau, Scikit-learn, PyTorch, Keras, Pandas, NumPy
• Utilized BeautifulSoup for web scraping, compiling a high-quality dataset of over 100,000 data points with 99% data integrity
• Created an interactive Tableau dashboard for real-time sales trend exploration, boosting decision-making efficiency by 75%
• Developed machine learning models (RNN, Random Forest, LSTM) in Python, with LSTM achieving RMSE of 0.05 and R2 score of 0.98
Advanced driver assistance system (ADAS)
Technologies used: C++, OpenCV, Nvidia jetson Nano
• Generated multicore embedded system using Nvidia jetson Nano 2g developer kit
• System assists drivers in the lane, traffic light, stop-sign, and pedestrian detection
• Successfully implemented AMP (Asymmetric multiprocessing) and SMP (Symmetric multiprocessing) using C++ and OpenCV which improved performance by 100% and obtained an average of 40 FPS
• Used HOG Descriptor, SVM model, ROI, Masking, Regression, Edge detection, Slope calculations, and Image preprocessing to improve driver safety and obtained an F-measure of 0.96
Gold Chase Game
Technologies used: C++, Python, OpenCV, Multithreading (Pthread), Makefile, Embedded Systems, Computer Vision
• Created multiplayer console game application- using C++, vagrant, distributed programming, signals, pipes
• Implemented semaphores for synchronization, shared memory to coordinate multiple processes, Signal handling, and message queues for automatic screen updates and chat functionality
• Aiming to complete the project within 4 weeks and reduced the deployment time by 50%
Netflix Data Analysis Dashboard
Technologies used: BeautifulSoup, Python, Tableau
• visualized Netflix's content library, offering insights into movies and TV shows available on the platform
• Created an interactive Tableau dashboard for analysis of content distribution by country, ratings, genres, and more
Fake News Detection using NLP
Technologies used: Natural Language Processing (NLP), Scikit-learn, Python, NumPy, Pandas, Matplotlib, Scatterplots, Data Preprocessing
• Developed a set of prediction models with python using classification algorithms like passive aggressive, gradient boost, logistic regression, and support vector
• Utilized vectorization for textual data which increased the computational speed and achieved an accuracy of all models above 94%
• Performed data visualization using seaborn, matplotlib, scatterplots
Maximum flow in the network
Technologies used: C++, Java, Python
• Analysed and implemented the Edmonds-Karp algorithm, Dinic's algorithm, and Push-relabel algorithm using highly optimized C++, Java and Python code
• Performed and compared asymptotic analysis of the algorithm's run time
• Generated test cases using real-world and synthetic data of different sizes and different complexity
Inventory management system for home
Technologies used: MongoDB, React, NodeJS, Express, Django, Python, GCS, AWS
• Full stack web development using two stacks- MongoDB/React/NodeJS/Express, Django/Python
• Cloud providers- GCS and AWS
• Deployed each stack within Docker images and provided a load balancer on the back-end
• Differentialized the performance and accessibility of resources by working on both stacks individually
• Developed CI/CD pipelines to automate the software delivery process and ensure seamless integration of code changes in the project
Custom kNN algorithm
Technologies used: Python, seaborn, matplotlib, scatterplots
• Built custom kNN algorithm from scratch in Python without using any external ML libraries
• Tested kNN on classifier and regressor datasets (Sklearn) and acquired 95% accuracy
• Achieved 100% efficiency as that of Sklearn’s kNN and visualized the results with matplotlib, seaborn, and scatterplots
Diabetes Detection System
Technologies used: Python, Swift, XCode
• Worked in a team to develop Android and iOS applications for the detection of blood sugar levels (diabetes) by scanning the human retinal image
• Developed image recognition model using Python (Sklearn) for android application and Swift, XCode for iOS application
• Trained model by providing own dataset and improved accuracy to 99%
E-commerce Application for online shopping
Technologies used: HTML, CSS, JavaScript, Django, Python, AWS
• Designed and implemented a user-friendly e-commerce interface using HTML, CSS, and JavaScript
• Created back-end functionality for user authentication, product management, and order processing using Django and Python
• Integrated payment gateway APIs like Razorpay for secure online transactions
• Deployed the application on AWS, conducted testing and debugging, and collaborated with team members providing regular updates and feedback on the project's progress and goal
Social Media website
Technologies used: MongoDB, React, NodeJS, Express, Bootstrap, Django/Python, AWS, GCP, SQL
• Constructed a social media web application using MongoDB, React, NodeJS, Express, Bootstrap, and Django/Python
• Hosted and deployed the application on cloud providers like GCS and AWS, ensuring high availability and scalability for worldwide users
• Built RESTful APIs to enable seamless communication and data management between the client-side and server-side components
• Migrated database from SQLite to SQL to enhance performance and scalability without impacting user experience
• Troubleshot end-user issues and led technology projects to enhance web application workflows and meet functional requirements and strategic goals
Chess Game
Technologies used: HTML, CSS, JavaScript, Django, Python
• Generated a Chess game web application using Python and Django framework
• Utilized HTML, CSS, and JavaScript to create an intuitive and visually appealing user interface
• Implemented Chess movement functionality using Django Forms and Models, with form validation that reports errors to the user
• Created a per-user board model, allowing each logged-in user to play a game independently, saving their progress in the database
• Conducted thorough research on user needs and preferences, resulting in a user-friendly interface and a 20% increase in user engagement
Organizer Website
Technologies used: HTML, CSS, JavaScript, Bootstrap, GCP, Django, Python
• Implemented a full-stack web service using HTML, CSS, JavaScript, Bootstrap, Django, and Python
• Formulated a personal to-do list feature (Tasks) and expense tracker feature (Budget) with add, edit, delete, and toggle completed state functionalities
• Utilized chartist.js library to create pie charts and bar charts to display summarized data
• Deployed Dockerized applications to Google Cloud Platform for remote accessibility
Food pantry management application
Automated testing, Jest, PostgreSQL, React, Docker, Agile methodologies, Git, CI/CD, TypeScript
• Updated food pantry management project (NodeJS to React and MongoDB to PostgreSQL) with Docker and automated testing
• Implemented a process to approve each backlogged Dependabot PR independently to ensure updates don't cause regression or other issues
• Added CI workflow for testing all pull requests and main branch with coverage reports and status badges to ensure continuous integration and delivery
• Led the creation and execution of comprehensive test plans, resulting in a reduction in bugs by 40% and an overall improvement in product quality