About Me

Hi! I'm a recent computer science graduate from the University of California, San Diego. I'm extremely passionate about learning and applying cutting-edge technologies towards real-world issues; I am particularly interested in working with artificial intelligence and data, as well as the surrounding infrastructure. I hope to become a productive engineer in roles where I can grow and learn from experienced team members and draw project experience.

Resume

Or, scroll down to the experiences section below.

Harsh Gurnani

LinkedIn

You can also find this, as well as other links, at the bottom of the page.

Harsh Gurnani

Projects

Harsh Gurnani

ALUM Mobile App

2023

Building an iOS mobile application (ALUM) to connect high school alumni with current students using Swift UI, MongoDB, Express, and Node.js; worked full stack on MVVM architecture and RESTful API routing. Increased user interaction with application and improved app performance by creating an asynchronous automated notification system using Firebase tools (Cloud Messaging) and Apple Push Notification Services.

Harsh Gurnani

Diabetes Prediction

2023

Applied classification methods to patient physiological dataset to predict presence of diabetes with 93% accuracy. Trained XG Boost, Random Forest, and SVM models and compared performance against logistic regression. Optimized model hyperparameters based on custom error rate to minimize prediction of false negatives.

Harsh Gurnani

PAP Inventory

2024

Built an inventory processing web application for Patriots and Paws, a non-profit organization providing furniture to veterans, using React, Node.js, Express.js, and MongoDB. Designed authentication system using Firebase Cloud Tools. Optimized MongoDB database queries to allow filtering and sorting of inventory form submissions.

Experience

Professional

Databricks

Software Engineering Intern

Developed a cluster-scheduling simulator to run scheduling-algorithm regression tests, and execute capacity-projection studies to quantify 35% compute cost reduction of migrating AI workloads onto microVMs.

Conducted 40+ simulation experiments across AWS/Azure/GCP, showing 95% fidelity to production behavior.

Built MCP tools, Claude skill, and agentic pipeline to simulate a cluster and generate practical metrics/visuals.


Amazon

Software Development Engineer Intern

Designed and implemented end-to-end system architecture for an internal AWS Bedrock/LLM powered chatbot to automate ~500 monthly support tickets about benefits, cutting average response time from 1-2 days to 30 seconds.

Engineered real-time RAG pipeline for AI agent to obtain updated user data, reducing hallucinations by 40%.

Used GPT-4o to deploy a load balanced Websocket API endpoint hosting the AI agent and handle traffic spikes.


Arista Networks

Software Engineering Intern

Engineer multicast routing solutions to enhance network performance and support cloud traffic patterns at scale.

Led refactoring of PIM multicast protocol code to integrate a new hashing system, improving efficiency by 20%

Developed Python and C++ tools on Linux platform for 10+ bug fixes, impacting 5,000 enterprise customers


Bear River Associates

Machine Learning/Software Intern

Utilized spaCy Named Entity Recognition NLP method to extract information from mail addresses; deployed on Amazon SageMaker and AWS API Gateway to develop scalable service and handle increased load.

Created a browser-based mail scanning interface in Angular using a cutting-edge document scanning SDK to expedite mailroom processing time by 50% and simplify service implementation on client-side devices.


Rose-STL Lab

Undergraduate Researcher

Conducting time-series classification using predictive LSTM model to classify human movement type.

Predicting causes for lower back pain by training DNNs on human kinematic and muscle activation data.

Executed data preparation (cleaning, normalization, etc.) of raw data to ensure quality input for deep learning.


Triton Software Engineering

Engineering Manager

Managed 8 developers to build custom applications for nonprofit clients, increasing target audience reach by 2x.

Built a Slack-based retrieval agent for GitHub and internal documentation, handling 100 queries per week from engineers.

Developed a shared Firebase authentication library adopted across multiple client projects, saving 5 engineer hours weekly.


Leadership

University of California, San Diego

Advanced Data Structures Tutor

Teach implementations of advanced data structures (BSTs, Graphs, Tries, etc.) in C++ in a fully remote course.

Facilitate student learning by holding lab hours to help students debug with GDB and draw DSA visualizations.

Assist instructor and TA's in testing programming assignments, assessments, and answering questions on EdStem

Skills

Contact Me

University Email: [email protected]

Personal Email: [email protected]

Phone Number: (925)-789-8897