Projects & Hackathons

Can be found at my GitHub: github.com/srikarpisupati

FinWise — A Portfolio Manager, Software Engineer        Sep 2024 - Dec 2024

• Developed Flask-based REST APIs to deliver investment portfolio analysis using FinGPT Dow30 model (PEFT)
• Fetched Yahoo Finance data (stock data, news) and created prompts to generate stock analysis & predictions

Fair Share — Undergraduate Research, Software Engineer        Jan 2024 - Dec 2024

• Researched fair-division algorithms to optimally allocate indivisible resources, given individual preferences
• Developed a user-friendly iOS app (Swift, Cloud Functions) to compute allocations, maximizing Nash Welfare
• Implemented SciPy’s LP and Min Weight Bipartite Matching algorithms, tested performance with synthetic data
• Used Firestore (NoSQL) to store users, devices & groups. Supported group membership using invitation codes.

CrowdCompute — HackIllinois Hackathon        Feb 2024

• Built a web application using React, facilitating the training of ML models on remote (Docker) machines
• Responsible for backend data storage using Firebase, managing host machine metadata, model files & weights

ILline — BuildIllinois Hackathon        Nov 2023

• Developed an iOS app to support device proximity detection, distance-based ranking, and notifications.
• Used CoreBluetooth and P2P pairing to automatically detect nearby devices, and assigned “queue” positions.
• Won 4th place out of 36 teams and 200+ contestants

American Sign Language Classifier        Oct 2023 - Nov 2023

• Trained and tuned a CNN based ASL classification model using the Sign Language MNIST dataset
• Created a Python Gradio web app to capture camera image and classify it using the pre-trained ASL model

Assignments

• Built core system utilities in C: custom malloc, concurrent client –server system, parallel make, and a basic shell
• Gradient Descent: Analytically implemented and compared Adam, SGD, and AdaGrad optimizers
• ML Projects: Trained and compared Random Forest, Decision Tree, kNN, and SVM models on UCIMLRepo data
• Implemented VectorDB search; Trained with PyTorch: k -NN, k -means, regression, CNN, and ResNet models
• Built regression model for predicting Revenue-Passenger-Miles from 5+ factors, for Kaggle Air Traffic dataset
• PhotoMosaic: Implemented k-d tree to map image sections to nearest tile by color, to build a photo mosaic
• IlliniBook: Developed a social network to search related people, shortest path, and connected components (BFS)
• Created a skill–job matching web app: designed DB schema & SQL queries (top companies, in-demand skills).

Awards & Leadership