Chittebbayi Penugonda

• Developed two Agents in Python with Google ADK: a natural-language-to-SQL engine for device search and a fault-analysis tool that identifies defective clusters and suggests root causes for investigation. • Led every phase of development, including research, experimentation, mock-data creation, design-document authoring, integration, and automated testing • Engineered backend services with Python and front end code with Kotlin and Typescript to integrate diagnostic agents into the existing platform, accelerating device-cluster analysis and delivering deeper insights.

- Addressed the problem of class imbalances (lack of data points for abnormal classes) in training datasets for Metabolomics prediction models - Researched a variety of methods that leveraged generative AI to create synthetic tabular data - Applied generative model as an alternative to the state-of-the-art for oversampling (SMOTE) by using iterative LLM fine-tuning with Pytorch

Network and Machine Learning Lab | Financial Services Innovation Lab - Employing FinBERT NLP model to perform financial sentiment analysis on Twitter data with Pytorch - Optimizing analysis accuracy by expected 3% with Retrieval Augmented Generation using dataset compiled and formatted with SQLite

- Performing statistical analysis (t-tests, histogram visuals, etc.) with Python libraries and GIS tools across multitude of data sources to analyze mass power outage failure in relation to social vulnerability - Increased geographic spatial data point granularity to find discrepancies of up to 22% in original paper’s county-wise social vulnerability assumptions

• Integrated project IDX (online IDE) into Google Maps Platform API samples using Angular framework, enabling users to seamlessly sandbox APIs, increasing customer engagement with GCP’s mobile APIs by 2.5% • Developed backend functionality to automatically enforce restrictions on newly created API keys, increasing security and reducing billing disputes • Undertook full development process, including analyzing code base, authoring design documents, implementing & reviewing code, writing testing scripts, and monitoring client-side events with SQL scripts • Coordinated with UX team, Cloud Credentials team, and fellow intern to engineer project specifications

- Tasked with creating an interface to streamline employee information retrieval and management for the HR team, currently used to manage 150+ employees - Created dashboard with React.js/Node.js using Firestore RESTful API to communicate securely with NoSQL database - Implemented Cron Jobs in JavaScript server to periodically check employee records and send expiry warning emails using Gmail API. - Developed secure Google Auth authentication and protected site-routing system that incorporated a hierarchy of user levels

- Led a team designated for data manipulation and analysis, implementing Agile workflow methodology - Used Selenium web scraping to gather current data on appliance market availability - Implemented a system in which users could upload pictures of appliance asset tags to a cloud database, and processed the images with Google Vision AI using Python - Utilized the Python Pandas library to analyze scraped and user-uploaded data, tailoring suggested services to maximize user convenience

Relevant Coursework: Data Structures and Algorithms, Machine Learning, Systems and Networks

Rank: 3 out of 588 students GPA: 4.7