Hi there - Nice to meet you!

I am Matthew Tracy.

I am an Applied AI Engineer, Builder, and Language Learning Nerd!

Feel free to contact me via email or chat with me now at MatthewAI!

About

Hi! Nice to meet you!! My name is Matthew Tracy. I graduated from UNC Chapel Hill with a BS in Business Administration and a minor in Data Science, and I am currently pursuing my MS in Analytics (Computational Data Analytics) at Georgia Tech with a 4.0 GPA. In college, I founded a nonprofit 501(c)(3) that taught hundreds of students about STEM, the arts, and physical fitness, while providing tens of thousands of dollars in support to my NC community.

Currently, I am a Senior Associate in the Data Analytics and AI practice at PwC, where I build AI-driven solutions and ML models for major healthcare organizations. My projects have contributed to $1.88M of a $2.2M client impact and helped secure a $3.6M contract extension.

My passions include entrepreneurship, building side projects, playing sports (tennis and soccer), and learning languages (Spanish and Chinese)!

Profile Image

Work Experience

Applied AI Engineer Senior Associate – Data Analytics and AI Practice

PricewaterhouseCoopers (PwC) | Raleigh, NC – January 2024 – Present

  • Helped grow an AI incubator into a full-scale AI & Automation Program within a leading healthcare organization, building strong trust with the C-suite. Projects contributed to $1.88M of a $2.2M impact and helped secure a $3.6M contract extension
  • LLM Agent System: Built an AI agent system using LangChain ReAct, FAISS, and AWS services to identify high-impact risk-adjustment records with batch OCR, multi-step LLM reasoning, embeddings search, and frontend UI, leading to a $1M ROI pilot and $4M annualized impact
  • LLM-Based Document Processing: Designed and deployed a large-scale text-LLM extraction and NLP pipeline using AWS (S3, Textract, Lambda, Bedrock, LangChain), reducing processing time from 12+ months to days. President shouted out the use case company-wide
  • XGBoost Model Development: Built an XGBoost model for a payer with 700k members, saving $1.9M annually through improved subrogation and payment recovery, integrating the model into the payer's existing claims workflow
  • Cost Reduction Analysis: Analyzed patient claim data using SQL and Python, identifying inefficiencies and eliminating ineffective vendor services, resulting in $800k in annual savings
  • AI Pod Leadership: Led an AI development pod with three team members, collaborating weekly with C-suite executives and healthcare clients to showcase results, resolve blockers, and refine strategies
  • LLM-Driven Pharmacy Policy Tool: Built a secure workflow tool using LLM agent and RAG to compare policy document versions, detect changes, and generate structured summaries. Reduced review time from 2.5 hours to 5 minutes

Nonprofit Advisor

Activate Kids 501(c)(3) | Raleigh, NC – October 2023 – February 2024

  • Led the strategic merger with Activate Good, creating a unified entity with over $840k in combined budget and a team of 7 full-time staff. Successfully integrated missions and operations, enhancing program efficiency, stakeholder engagement, and community impact

President, Founder, Chairman of Board

Cooking Up Character 501(c)(3) | August 2021 – October 2023

  • Orchestrated the operations of a board of six industry professionals, delivering service-learning programs in public schools and parks to 700+ youth, amplifying community impact by $70k
  • Developed and implemented innovative service-learning initiatives, empowering youth to become proactive changemakers, increasing program participation by 27% year-over-year
  • Strategized and implemented innovative problem-solving frameworks for youth, driving a 20% year-over-year surge in the organization's community impact

Technology Consulting Summer Associate

PricewaterhouseCoopers (PwC) | Boston, MA – June 2022 – August 2022

  • Led B2B customer interviews, deriving critical insights used to code and develop Python-based data analytics models utilizing 300k data points, directly influencing strategic cost reduction decisions adopted by Executive VPs
  • Developed a PowerBI dashboard to monitor 70 KPIs, enhancing vendor productivity and system reliability significantly

Business Analyst Intern

Vita Safety Partners | Raleigh, NC – June 2022 – August 2022

  • Supported clients by conducting comprehensive research and analysis of the emergency services market
  • Enhanced a CRM database encompassing over 10,000 leads, catalyzing a 15% improvement in lead conversion rates
  • Analyzed multi-platform data to create tailored marketing strategies, boosting campaign effectiveness by 10%

Strategy Consultant

Nike | Chapel Hill, NC – January 2022 – August 2022

  • Developed and pitched Nike's 'Like New' circular economy initiative, allowing customers to exchange lightly used apparel for credit, fostering a closed-loop system that enhances sustainability and expands market reach through recycling

Technical Projects

Browser Adaptation Challenge Agent

January 2026 – February 2026 | Demo Video

  • Built a fully deterministic Playwright agent that solves all 30 sequential browser automation challenges (click, scroll, drag-drop, canvas, audio, video, WebSocket, shadow DOM, iframes, etc.) in under 3 minutes with zero LLM calls
  • Engineered React fiber tree traversal to extract component props and call internal callbacks directly, bypassing an intentionally unsolvable recursive iframe challenge with an off-by-one depth guard
  • Developed a second hybrid variant using Claude Haiku vision as a fallback for math puzzles and ambiguous challenges, completing all 30 steps in 297.7 seconds with 7 targeted LLM calls
  • Implemented obstacle dismissal (popups, overlays, fake buttons, confetti), stale state detection with forced React remounts, and challenge-type classification via DOM selectors -- no sessionStorage decryption or external services

TennisNearby.com

December 2025 – Present

  • Built a modern tennis court discovery platform across the Triangle and broader North Carolina with interactive map/list/split views, "Near Me" search, and advanced filters (amenities, access type, surface, cost, lights, bathrooms, etc.)
  • Designed and implemented a geospatial backend using Supabase (PostgreSQL + PostGIS) to support proximity queries, distance-based sorting, and scalable court search
  • Developed core product features including SEO optimization, favorites and sharing, and foundations for community features to help users find and connect with other players

SacredStatistics

February 2024 – May 2024 (Paused)

  • Built an end-to-end pipeline to collect, clean, and consolidate Catholic saints data into structured JSON/CSV datasets for analysis and scalable reuse
  • Prototyped a BeautifulSoup-based scraper, then migrated to an LLM-driven structured extraction approach using Perplexity API with few-shot prompting to enforce a strict JSON schema
  • Implemented validation and batch processing to handle edge cases, improve consistency, and generate clean datasets for downstream EDA and applications

Want to Chat with an AI Matthew Right Now?

Talk with my AI me - MatthewAI!

Talk with MatthewAI
Next steps?

Don't be a Stranger!

Feel free to drop me an email if you want to connect!
Thank you very much! :)

Get in Touch