Machine learning student at CMU.
Engineer and researcher.
Machine learning · AI systems · Research
I’m an M.S. Machine Learning student at Carnegie Mellon University and a graduate student researcher in the Human Sensing Laboratory. My current research explores how large language models internally represent positions on social and political issues.
Previously, I built basketball video analysis systems at Paloa, worked on software and infrastructure at Cvent, and co-founded SiliconMCP, formerly LoopCell, to support AI driven chip design. I graduated from UT Austin with an Honors degree in Computer Science, summa cum laude.
Outside of work, I enjoy public speaking, nonprofit fundraising, sports, reading, cooking, and traveling. My past modeling and acting work included representation by Entourage Model and Talent Agency (opens in a new tab).
Current questions and earlier investigations.
September 2026 to Present
Graduate Student Researcher
Human Sensing Laboratory, Carnegie Mellon University
Developing a metric to audit how large language models internally represent positions on value-laden social and political issues. I evaluate consistency across prompts and contexts using open-source models, informed by interpretability and value-measurement research.
December 2024 to July 2025
Independent Research with Dr. Anshumali Shrivastava
ThirdAI
Designed retrieval benchmarks covering single-document retrieval and multi-document questions with single or multiple answers. I curated datasets, measured retrieval quality and latency, and evaluated eight transformer NER models on precision, recall, F1, and CPU inference latency under 10 milliseconds. I also tested self-hosted vision-language models for document processing and wrote reports informing architecture decisions.
February to July 2024
Independent Research with Prof. Chandrajit Bajaj
The University of Texas at Austin
Compared genetic algorithms, deep reinforcement learning, and graph neural networks for dynamic network optimization. I documented data flow, equations, and optimization trade-offs, and proposed a hybrid reinforcement learning and GNN framework for scalable deployments.
June to August 2020
Research Fellow
The University of Texas at Austin
Designed a Raspberry Pi scanner and Python tools to discover medical devices on a healthcare network, classify vendors, and generate a security inventory with IP addresses, MAC addresses, and device details. I also explored network fingerprinting to improve detection accuracy.
July to December 2017
Student Researcher with Dr. Timothy Riedel
DIY Diagnostics Lab, The University of Texas at Austin
Built an Arduino recording device to capture mosquito wing beat audio, then used noise removal and frequency analysis to study species-specific acoustic signatures.
AI systems, products, and experiments from recent years.
AI safety · Uncanny Valley Fest
September 2026
Won Best HACK + CREATE at the Uncanny Valley Fest AI filmmaking hackathon. Built a live evaluation environment in under four hours, with an LLM agent, an interactive guard, and a judge that scores deception, oversight subversion, and self-preservation. Runs support streaming, recording, replay, and an offline mode, backed by 33 tests.
Node.js · Express · Python · OpenCV · World Labs Marble · Tavus
Computer vision · HackCMU Food Track
September 2026
Built a native iOS game that turns a photographed meal into a fighter whose stats reflect its nutrients. A YOLO segmentation pipeline identifies foods, estimates portions, and maps them to USDA nutrition data within a four-second budget. A shared deterministic simulation keeps Swift and Python fight logs aligned for live multiplayer battles.
SwiftUI · Python · FastAPI · YOLO · WebSockets · Redis
Co-Founder & CEO · AI for chip design
July 2025 to September 2026
Architected Model Context Protocol infrastructure for AI driven chip design, including evaluation-as-a-service for LLM-generated SystemVerilog and agentic RTL and testbench generation. Led technical direction and simulation-based evaluation workflows for a pilot with a leading frontier model company. The team placed in the top 10% of Y Combinator applicants.
MCP · SystemVerilog · RTL generation · Simulation · Agent orchestration
Founder · LLM applications
March to December 2024
Built a platform that turns static resumes into personalized, shareable chatbots, enabling real-time questions about a candidate’s experience and skills. Filed a provisional patent application for distributing customized large language model chatbots.
Python · Streamlit · MongoDB · AWS Lightsail · LLMs
Founder · LLM applications
October to December 2024
Built an AI hiring application that parses resumes, extracts skills, and ranks candidates against job descriptions. It produces recruiter-ready summaries and match scores, with a single-link submission workflow for applicants.
Python · Streamlit · MongoDB · AWS Lightsail · LLMs
Data science · Sports
May 2022
Cleaned and explored baseball data, engineered features, and compared classification models for pitch prediction. The final support vector machine model achieved 93% accuracy.
Data mining · Feature engineering · Model evaluation
Data science · Sports
December 2021
Explored relationships between advanced basketball metrics and player positions through data wrangling, visualization, classification, and prediction. The project classified whether a player was a guard with 71% accuracy.
Data analysis · Visualization · Classification
Deep learning · Computer vision
December 2021
Built a neural-network agent for 2v2 ice hockey using convolution, batch normalization, and ReLU layers to predict a puck-location heatmap. The agent beat 80% of the competing AI agents.
Neural networks · Computer vision
Machine learning · Natural language processing
May 2021
Built a BERT-based model to identify named entities in tweets, using a linear layer and softmax over per-token representations. The Natural Language Processing final project achieved over 70% accuracy.
Python · BERT · NLP
Full stack development
March 2017
Built a responsive volunteer signup and scheduling application for Austin libraries, with administrator check-in and automatic reporting.
Node.js · MongoDB · Bootstrap · jQuery · Heroku
Android · Research
July to August 2016
Developed an image-pattern eye-testing application using an external ultrasonic sensor through UT Austin’s Research Methods Program under Dr. Ruth Shear.
Android · Sensors
Full stack development
June to November 2015
Built a responsive adoption application for Austin Animal Shelter with intelligent search and notifications covering more than 500 dogs.
Node.js · MongoDB · Bootstrap · jQuery · Heroku
Android · Robotics
June to August 2015
Used a smartphone’s magnetic orientation sensor to steer a Mindstorm robot in a straight line, interfacing with an external motor controller over Bluetooth.
Android · Bluetooth · Mindstorm
Built a VLM-first basketball analysis pipeline for player identification, box scores, and game summaries, with RF-DETR, ByteTrack, jersey OCR, and SigLIP. Created Paloa BENCH against 469 ground-truth shots across four games. A QLoRA fine-tuning and evaluation workflow improved shot-detection accuracy from 79% to 87%. Also shipped a five-service Node/Express AI gateway with authorization, rate limiting, cost attribution, and Supabase analytics for the coach-facing app.
Twelve Labs Pegasus · Qwen-VL · Modal · YOLO · Supabase
Architected a cross-service optimization that reduced latency by 35% and built Datadog monitoring that halved incident recovery time. Integrated AI development frameworks that doubled team productivity and joined a two-person, principal-engineer-led feature initiative impacting more than 10,000 users.
Software architecture · AI development workflows · Datadog
Built React and Redux components that improved dashboard responsiveness by 15%, RESTful APIs serving more than 1,200 clients, and WebSocket handling for over 500 messages per second. Optimized rendering, database consistency, memory use, SQL procedures, and test coverage.
React · Redux · AG-Grid · Spring Boot · PostgreSQL · Couchbase
Led two weekly CS 314 Data Structures sessions and designed lesson plans around the course material.
Teaching · Data structures
Developed REST endpoints for a service impacting over 21,000 users, improved SQL performance by more than 30%, and built a React application launched to 1,200 early adopters, with more than ten unit and integration tests.
Spring Boot · PostgreSQL · React · Mockito · Karate
Samsung Electronics
June to August 2021
Built a Python power-analysis automation suite that reduced report generation time by 40%. Cut a manual analysis workflow from three to four hours to 45 minutes while working on pre-silicon GPU power measurement and optimization.
Python · GPU power analysis · Automation
Supported Computer Architecture through office hours, weekly discussions, grading, and daily practice problems.
Teaching · Computer architecture
Led the mobile DevOps pipeline for Protect Texas Together, downloaded by more than 10,000 people. Built user-facing functionality with a ten-person team of professors and students.
React Native · Django · Fastlane · Bitrise
Built a cloud reporting system using data for over 800,000 artists, with REST APIs, AWS SQS, microservices, SQL, and an S3 report interface. The system performed 50% better than its predecessor.
Java · Spring Boot · AWS SQS · AWS S3 · SQL
Panera Bread
February to May 2018
Supported customers at the front desk and register and helped with food preparation and end-of-day operations.
Customer service
Python
Java
HTML
CSS
JavaScript
SQL
C
C++
MATLAB
MongoDB
MySQL
PostgreSQL
Shell
Git
AWS
Linux
React
Bootstrap
Django
Express
jQuery
Node.js
Spring Boot
REST APIs
Spark
Hadoop
PyTorch
TensorFlow
Jenkins
Maven
Gradle
Docker
HerokuAdditional tools include WebSockets, AG-Grid, Streamlit, Couchbase, Datadog, Splunk, Hadoop, Linux, Jenkins, Maven, Gradle, Bash, JIRA, Arduino, SOX, Audacity, and SolidWorks. Project-specific work also includes Swift, SwiftUI, OpenCV, SQLite, Redis, Auth0, and Railway.
M.S. Machine Learning
Expected December 2027 · Pittsburgh, Pennsylvania
Introduction to Machine Learning, Probability and Mathematical Statistics, Independent Study (Research), and Intermediate Deep Learning.
B.S. Computer Science with Honors, Summa Cum Laude
May 2022 · GPA: 3.96 / 4.0 · College Scholar
Neural Networks, Natural Language Processing, Computer Vision, Data Mining, Network Security, Big Data, Elements of Data Science, Probability and Statistics, Matrices and Matrix Calculations, Operating Systems, Computer Architecture, Computer Networks, Data Structures, Algorithms and Complexity, and Software Engineering.
November 2024 to Present
Board and Fundraising Committee Member
Helped raise more than $50,000 in donations and built a centralized database to consolidate information from multiple spreadsheets. Support outreach, grant proposals, volunteer engagement, and events.
July 2023 to Present
Leadership roles
Served in roles including Area Director, President, and Vice President of Public Relations. Supported a network of seven clubs and led a group of more than 30 people, helped grow membership by 25%, and contributed to President’s Distinguished Club recognition. President since June 2024; VP of Public Relations from July 2023 to June 2024.
January to November 2020
Finalist, interdisciplinary project at UT Austin
Helped design and prototype a virtual exhibition highlighting oral histories and portraits of artists of color in Austin and The Bahamas, collaborating on research, storytelling, and digital curation.
Fall 2016 to Winter 2018
Team Lead
Led a high school team to the world finals, with code deployed on the International Space Station. The team placed third among more than 200 teams.
Completed more than 60 AI and systems design courses through providers including DeepLearning.AI, DataCamp, and MongoDB University. Societies include ACM, ISSS, and Toastmasters. Languages: English and Spanish.
For an interesting problem, a creative idea, or a conversation, I’d love to hear from you.
