Bisti Potdar
About Me
Hi! I'm Bisti.
I am a Computer Science student at Purdue University, with a passion for AI/ML and building intelligent systems. I am dedicated to creating technology that makes a tangible impact.
In my free time I like pursuing content creation, trying new resteraunts, and reading!
Relevant Coursework
Completed
CS251: Data Structures and AlgorithmsCS250: Computer ArchitectureMA265: Linear AlgebraCS240: Programming in CCS176: Data Engineering in PythonCS180: Object Oriented Programming
In Progress
CS252: Systems ProgrammingCS211: Competitive ProgrammingCS471: Introduction to Machine Learning
Skills
AI / ML
GANsLLMsRAGAgentsPytorchRLDeep LearningNumpyPandasSQLNoSQL
Languages
C/C++JS/TSReactKotlinPythonJavaHTML/CSS
Misc
REST APIGitLinux
My Experience
May 2025 – Present
ML@Purdue Vice President of Outreach
Machine Learning at Purdue
- Organized CatapultHacks, Purdue's biggest AI and Entrepreneurship hackathon of 150+ participants
- Led outreach operations by recruiting new members, coordinating faculty partnerships, organizing technical workshops, and managing the organization's website and sponsor communication
September 2025 – Present
Project Manager
Purdue Momentum
- Managed a 5-member student team developing CodeTutor AI: an educational technology tool that leverages LLMs to explain compiler/runtime errors and suggest targeted fixes
- Oversaw project planning, sprint management, and cross-functional collaboration while integrating Python, Node.js, React.js, Docker, WebSockets, and the VS Code Extension API into a semester-long build
May 2025 – Aug 2025
Computer Vision Researcher
Purdue University AIM Lab
- Collaborated with a research team under Dr. Lu to design a mobile app improving string musicians' posture and bow hold through advanced computer vision and machine learning models
- Integrated on-device YOLOv11 bounding box models, enabling real-time feedback within a user-friendly mobile application
March 2025 – Present
Web Developer
InnovateHER Hacks
- Used React Native to develop the website for InnovateHer, Purdue's flagship women-centric hackathon
Projects
Goaly - Accountability Coach
October 2025
Built an MCP-compliant goal tracking server utilizing agentic workflows and LLM integration to decompose high-level objectives, leveraging multi-modal analysis to verify daily progress and enable AI agents to autonomously manage user accountability.
JavascriptPythonExpoGORESTful APIPoke.aiSupabaseRender
BudgetBuddy
June 2025
Developed a budget generation system that converts draft documents and templates into clean, auto-filled spreadsheets, reducing manual work and errors. Includes in-browser template editing via cell-to-text mapping using key value pairs.
TypeScriptNext.jsxlsmMongoDB Atlas
BetterBytes
February 2025
Created for Boilermake 2025, BetterBytes is a web application with in-built barcode scanning functionalities. Using computer vision and LLMs, we built a platform where users can make better choices on which foods to eat based on their own personalized food goals.
AIComputer VisionLLMsUX/UIWeb Applications
Eve. - Menstrual Product ScannerWinner
January 2025
Won first place at InnovateHER hackathon in the health & fitness category. Created a web app that scans menstrual products using the user's web camera and provides health and safety information about the product using the GROQ API. Aimed to solve the lack of transparency women face regarding menstrual products.
LLMComputer VisionCNNData AnalysisGROQ API
Research
GLEAM: GAN and LLM for Evasive Adversarial Malware
October 2023
Published paper in the 2023 14th International Conference on Information and Communication Technology Convergence (ICTC) division of IEEE, receiving first authorship, invited to present at the conference in Korea. Worked alongside a team to create a state-of-the-art machine learning model, coined "GLEAM," to test the creation of evasive malware for cybersecurity systems.
GANLLMMachine LearningCybersecurityIEEE Publication
AIM Lab: Computer Vision Research for String Musicians
May 2025 – Aug 2025
Collaborated with a research team under Dr. Lu to design a mobile app improving string musicians' posture and bow hold through advanced computer vision and machine learning models. Integrated on-device YOLOv11 bounding box models, enabling real-time feedback within a user-friendly mobile application.
YOLOv11Computer VisionMobile AppMachine LearningResearch
RED40GAN
2024
Built RED40GAN, a custom Generative Adversarial Network (GAN) using TensorFlow/Keras to generate synthetic UV-Vis spectrometry data for a Beer–Lambert Law calibration curve. Integrated machine learning with experimental chemistry to model Red 40 concentration, producing synthetic data that closely matched real lab measurements. Reduced experimental percent error from 376% to 38%, demonstrating how generative AI can improve lab accuracy and reduce human error in chemical analysis.
TensorFlowGenerative Adversarial Networks (GANs)UV/Vis SpectroscopyKerasMachine Learning