Available for opportunities

Haren Gannarapu

SWE @ Princess Cruises (Carnival) · Prev AI/ML Intern @ Oasis · CS @ UCF. I build AI-powered systems at the intersection of full-stack engineering and creative problem-solving.

9+ Projects Shipped
3.8 UCF GPA
20+ Technologies
About

Engineering at the edge of AI and human-centered design.

I'm an impact-driven Computer Science student at UCF passionate about designing AI-powered systems that bridge real-world utility with thoughtful design. I thrive at the intersection of AI/ML, full-stack engineering, and creative problem-solving — bringing precision and imagination to every project.

From building autonomous robot navigation systems to creating intelligent scam detection tools and fintech applications, I write clean, efficient code and design experiences that genuinely matter. My projects span AI research, full-stack development, and robotics — always with a focus on solving real problems for real people.

Python
JavaScript
TypeScript
React
Next.js
Node.js
Flask
PyTorch Geometric
AWS
Kubernetes
Docker
Terraform
OpenAI API
YOLOv5
Firebase
PostgreSQL
Prometheus
Grafana
Arduino
LiDAR
Tailwind CSS
Git
Resume

My credentials, in one place.

Projects

Things I've built.

🏆 Winner — HackaBULL VII 2026 Featured

Revive — Emergency CPR Training System

Next.js 16 · React 19 · TypeScript · Arduino UNO R4 Minima · Gemini 2.5 Flash · ElevenLabs TTS · Web Serial API · Tailwind CSS 4 · Vitest

Won HackaBULL VII by building a hardware-software CPR training system that turns $50 of Arduino hardware into a full cardiac emergency simulation. A force-sensitive pad streams compression peaks over Web Serial at 100Hz — Gemini 2.5 Flash generates live patient scenarios, simulates real-time physiology (HR, BP, SpO2, rhythm) based on your depth and rate, and coaches you via three distinct ElevenLabs voices (calm instructor, 911 dispatcher, panicked bystander) with sub-50ms correction latency. Your eyes stay on the patient. Fully scored against AHA protocol with a detailed session debrief. 222 Vitest tests. Triple fallback chain: pad disconnected → keyboard mode, Gemini timeout → cached scenarios, ElevenLabs down → browser SpeechSynthesis.

🏆 Winner — Hack The Bay 2026 Featured

ReplAI — Conversation Practice Platform

Next.js · TypeScript · React · Gemini 2.5 Flash · LiveKit · MediaPipe · Three.js · Tailwind CSS

Won Hack The Bay 2025 by building a real-time AI role-play platform for high-stakes conversation practice — salary negotiations, HR disputes, legal client intake. An AI plays the other person while MediaPipe tracks webcam body language (eye contact, posture, fidgeting) and LiveKit analyzes voice (pace, filler words, energy) simultaneously. For the Legal Intake scenario, a custom Levenshtein + Jaccard fuzzy-matching engine extracts named entities from live conversation transcripts and surfaces opposing-party conflicts against a firm database in real time — before the meeting is even over.

Featured

KnightMobile — Autonomous Robot System

Python · Flask · React · YOLOv5 · Arduino · iPhone LiDAR · WebSockets · D* Lite

Built an autonomous robot from scratch that navigates unknown environments using AI vision and spatial mapping. Integrated Arduino motor control with iPhone LiDAR for real-time obstacle avoidance and path planning. Connected Flask API + React dashboard for live telemetry and control with sub-250ms feedback latency. Improved navigation reliability by 40% vs ESP32 prototype through adaptive mapping and motor feedback loops.

Featured

CardFinder — Smart Card Analysis

Next.js · TypeScript · Python · CFPB API · Recharts · Zod · Vercel

Created a fintech web app benchmarking 300+ U.S. credit cards using CFPB data and scraped issuer information, producing a transparent scoring algorithm for informed decision-making. Built interactive sliders and analytics charts for real-time fee, APR, and reward comparisons. Collaborated in a 4-member team following CI/CD practices and type-safe API design with Zod validation. Deployed on Vercel for seamless user experience.

DocuLens — GemiKnights 2025

React · Node.js · TypeScript · Firebase · Google Gemini API · Docker · OCR

AI-powered tool helping immigrants identify tax form errors using OCR and natural language processing. Developed an intelligent system to parse and summarize U.S. tax PDFs using Gemini API for document parsing and error explanation. Implemented Firebase rules and audit logging to ensure secure document access. Featured PDF uploads, form detection, fallback logic, and rate-limiting for production-ready deployment. Major university-sponsored project.

ScamBuster AI + Chrome Extension

React.js · Node.js · Express · OpenAI API · Tesseract.js · OCR · Tailwind CSS

AI-powered scam detection tool that analyzes suspicious messages, emails, and websites to protect users from fraud. Built an OCR → LLM pipeline with schema validation, fallback logic, and timeout resilience achieving 95% uptime. Created an explainable red-flag panel combining GPT reasoning + regex severity scoring. Developed a Chrome extension for instant webpage analysis with debounced listeners and secure REST calls. Backend deployed on Render for scalability.

WeatherDash

Python · Flask · OpenWeatherMap API · Matplotlib · REST APIs

Dynamic weather dashboard with real-time data visualization and weather-responsive backgrounds. Features city-based weather search, interactive temperature/humidity charts, and intelligent background changes that adapt to current weather conditions using Flask backend and OpenWeatherMap integration. Demonstrates full-stack capabilities with seamless API integration and responsive design.

AI Tutor-Matching System

HTML · CSS · JavaScript · UX Research · UI/UX Design

Co-authored a 9-page research paper for an AI-powered tutor-matching system at UCF. Built an interactive UI demo simulating live tutor-student pairing with subject and availability input fields, responsive output panel, and frontend-only logic. Focuses on improving educational accessibility through intelligent matching algorithms.

Python IDLE Guide

Python · Markdown · Technical Writing · Documentation

Authored a comprehensive beginner-friendly installation and usage guide for Python's IDLE environment. Included detailed screenshots, simplified setup instructions, and troubleshooting section to help new programmers get started with Python development. Demonstrates technical communication skills and commitment to making programming accessible.

Experience

Where I've shipped.

May 2025 – Aug 2025 AI / ML

Software Engineering Intern — AI/ML

Oasis (AI Travel Startup) · Python, PyTorch Geometric, AWS, Kubernetes

Built ML infrastructure for an AI travel startup's recommendation engine — a heterogeneous graph neural network modeling users, posts, and tags to surface personalized travel suggestions at scale.

Architected a heterogeneous GNN recommendation system using PyTorch Geometric, modeling users, posts, and tags as an interaction graph; built the full PostgreSQL-to-tensor data pipeline for link prediction.
Re-architected AWS infrastructure with Terraform and Kubernetes IaC, reducing dev environment build times by 93% and fully eliminating manual data processing steps.
Built a model observability pipeline with Prometheus and Grafana and deployed Triton Inference Server with KEDA autoscaling for horizontal GNN inference scaling, cutting mean time-to-debug by 60%.
Nov 2023 – May 2024 Data · Education

Data Analyst & Academic Tutor

Kumon · Davie, FL

Leveraged data analytics to improve student outcomes and educational program effectiveness. Combined technical skills with educational expertise to create data-driven insights for student success.

Developed SQL dashboards analyzing attendance and performance trends for 70+ students, improving retention insights and program effectiveness.
Built early-warning reports identifying at-risk students 2 weeks sooner to support timely intervention and improve success rates.
Provided one-on-one tutoring in mathematics and reading comprehension, helping students improve academic performance.
Utilized data visualization tools to communicate insights to program managers and parents.
Contact

Let's build something.

I'm always interested in discussing new opportunities, innovative projects, and collaborations. Whether you're looking for a dedicated developer, AI researcher, or just want to chat about technology — feel free to reach out. I respond to all messages within 24 hours.

📧 gannarapuharen@gmail.com
📍 Orlando, Florida

Project Demo