Agentic AI · LLM · Automation

Minwoo Park

AI Engineer

Studying AI at Carnegie Mellon University.

Turning data into decisions, utilizing AI in a human loop.

Resume
Carnegie Mellon UniversityUniversity of GeorgiaGeorge Mason UniversityPuzzleDataAndersen CorporationAccuver America

Studied and worked at 6 organizations

scroll down↓

Experience

From telecom to sustainability to AI engineering, solving problems with data.

Data Analytics Engineer Intern

Accuver America (opens in a new tab)
90%+ faster reconciliationFeb 2026 - Jul 2026

Sole engineer on an AI expense-reconciliation web app for the whole 30-person US office, plus weekly-run internal tools that replaced manual finance and reporting work for the CEO, CFO, HR, and branch management.

PythonGeminiClaude

Sustainability Data Analyst Intern

Andersen Corporation (opens in a new tab)
$20K+ billing errors foundJun 2025 - Aug 2025

Built a Python audit pipeline for five years of vendor waste and billing data for the sustainability team, so a vendor file could be loaded and come back with its errors already coded.

PythonPower BIData Quality

Data Analytics & Process Intelligence Intern

PuzzleData (opens in a new tab)
+5% final-stage conversionMay 2024 - Jul 2024

Analyzed web and app visit logs and procurement event logs with process mining (ProDiscovery) to find funnel drop-offs and approval-route deviations.

ProDiscoveryPythonR

Skills & Technologies

The tools and frameworks behind my work.

Data Engineering

Building pipelines, audit systems, and schema migrations that hold up in production.

3M+

Records

729

PRs

5%

Conv. Lift

$20K+

Cost Found

FlaskSQLiteDuckDBPostgreSQL / SupabaseDockerGitHub ActionspytestPlaywright

AI Engineering

Building agentic, multi-agent AI tools (LLM orchestration, RAG, and cross-model review loops) to automate workflows and augment human decisions.

Multi-AgentRAGLLM EvaluationClaude APIGeminiOpenAIPrompt EngineeringMCPOCRText-to-SQL

Analytics

Turning complex datasets into clear, actionable insights through visualization and statistical analysis.

Power BIStreamlitProcess MiningTableau

Languages

Core languages for data manipulation, analysis, and reporting.

PythonSQLRJavaDAX

Certifications & Languages

AWS-certified and bilingual, with work experience across the US and South Korea.

AWS AI PractitionerAWS Cloud PractitionerLean Six Sigma Yellow BeltEnglishKorean

Education

The foundation behind every insight.

Carnegie Mellon UniversityAug 2026 - Dec 2027

Master of Information Systems Management

Business Intelligence & Data Analytics (BIDA) · Heinz College

Concentration: Artificial Intelligence Management

Pittsburgh, PA

Projects

Real-world problems, data-driven solutions.

Telecom RAG

5G-domain RAG prototype: Q&A with source citations over 7 telecom standards whitepapers (2,034 chunks, MMR retrieval, source and page citations).

PythonLangChain+3

Expense Reconciliation

Full-stack receipt-reconciliation web app for a corporate finance team: multi-provider vision OCR matched against card statements, running in production.

Python / FlaskGemini 2.5 Flash+4

Sharday

Collaborative 2-4 person travel planner (oursharday.com): React 19 + Supabase with realtime sync, shipped from PRD to launch as a personal project.

React 19Supabase+3

Golmok

Agentic job-matching pipeline with source-grounded verification: scheduled Claude agents judge each internship posting, and a separate run re-reads the source posting to check their calls before anything is recommended. Built on Google Sheets and Apps Script; every application is submitted by hand.

Apps ScriptClaude Cowork+3

Mangchi

Agentic code-hardening plugin: Claude writes and decides, Codex CLI critiques one axis at a time until convergence.

Claude CodeCodex CLI+3

ddaro

Worktree-based parallel workflow for Claude Code: isolated branches, crash recovery, and CI-orchestrated merges.

PythonGit Worktree+3

Claude Code Plugin Suite

Open-source agentic tooling for Claude Code (7 plugins: batchim, prism, triad, mangchi, ddaro, boda, and prism-devil, since absorbed into prism), built at Accuver, used on every PR there, and handed to the successor engineer.

Claude CodeMulti-Agent+3

Walmart Purchase Prediction

Regression model comparison in R on 550K+ Black Friday transactions: linear, ridge, lasso, random forest, and GBM. Log-linear regression came out best (R² 0.74).

Rcaret+3

Procurement Process Mining

Process mining on a procure-to-pay system: purchasing, approval, and settlement logs joined into one event log, then discovery and conformance checking in ProDiscovery.

ProDiscoveryPython+3

Ready to connect?

Open to collaborations, data challenges, and new opportunities. Let's build something meaningful together.

EmailLinkedIn

[email protected] · Pittsburgh, PA · Resume (PDF)