Joshua
Vallabhaneni

Built a system that converts design mockups into production-ready code. Combines CV, RL, and agent-based reasoning to explore code variants and evaluate them via a custom similarity metric. Leveraged NVIDIA H100s on CoreWeave to cut rendering time from 20 -> 5 minutes, outperforming v0 and bolt by 5% in visual similarity.

Built a smart routing system that optimizes LLM API usage for businesses/users. The product analyzes incoming prompts, predicts response complexity and length for different LLM models, and automatically routes queries to the most cost-effective model while maintaining output quality.

Developed a next-generation disaster recovery platform that empowers wildfire-stricken communities to rebuild faster and smarter. FireSync uses AI for intelligent volunteer-task matching, real-time coordination, and impact tracking.

Engineered full-stack web app using React, Flask, MongoDB featuring custom-made automated matching algorithm, Google/LinkedIn API, NLP-based mission alignment scoring, and email automation.

Engineered a Chrome extension for 3FA using Python, FastAPI, and MongoDB Atlas, incorporating real-time liveness detection, facial recognition, and dynamic hand gesture recognition.

Developed a Flask-based application that harnesses the Open-Meteo API to predict flight delays by analyzing 5 real-time weather metrics. Engineered a RandomForestRegressor model using pandas and scikit-learn.

Employed JDBC to connect to a MySQL database and utilized SQL queries to analyze the efficacy of the drug Tagrisso. Executed OLS Multiple Linear Regression using Apache Commons Math for model training and predictive analytics.

Developed a system for measuring multiple vitals, storing results, and displaying them in real-time through a web portal and mobile application. Features a custom-built AI-based advice engine for basic health diagnostics.

Created an automated matchmaking software that has efficiently connected over 700 users. Employs algorithmic techniques via Google Apps Script to foster resident learning hubs hands-free.

Developed a device to prevent hot car child deaths by detecting a baby in a car seat and sending automated alerts when the car's temperature crosses certain thresholds. Features a multi-level alert system.