Where I've worked

1. AI Engineer

   Aug 2026 – Present

   Johnson & Johnson · Raritan, NJ

   • Architected an end-to-end Databricks pipeline (Oracle to curated Delta tables to S3) powering an AI-driven complaint scoring system processing 30,000+ complaints annually, deployed via Jenkins CI/CD across dev/qa/prod.
   • Engineered an LLM-based scoring orchestration layer with batch processing, prompt caching, and checkpointing, cutting compute costs 50% and saving SMEs weeks of manual review annually.
   • Extended a shared Python-based AWS Bedrock AgentCore runtime with configurable model selection across the full request chain, giving a 55-member team flexibility to experiment with new AI tooling.
   • Built and deployed a Hugo static site in Go across 4 environments with Jenkins CI/CD, and vendored the Mermaid.js diagramming library to remove a runtime dependency, accessible company-wide.
   • Enforced explicit JSON output contracts and streamlined orchestration routing across LLM scoring modules, cutting processing costs 70% and hardening prompts against injection risks.
   • Python
   • Databricks
   • Delta Lake
   • AWS Bedrock
   • LLMs
   • Prompt Engineering
   • Jenkins
   • CI/CD
   • Go
   • Hugo

2. Software Development Intern

   Jun 2026 – Jul 2026

   Re:Members · Remote

   • Engineered member-portal features in Elixir/Phoenix LiveView and Ecto/PostgreSQL, including database-scoped email-template access control and a config-driven avatar column, serving ~80,000 members across 45+ organizations.
   • Diagnosed and resolved a Stripe SetupIntent state-sync bug in Checkout (Phoenix LiveView), eliminating a false "payment blocked" state and safeguarding a $500K client renewal.
   • Authored 80+ ExUnit/Phoenix.LiveViewTest regression tests via test-driven development, tracked with ExCoveralls and enforced in a containerized GitHub Actions CI/CD pipeline, protecting member workflows from regressions.
   • Eliminated a critical member-facing form crash spanning 3 distinct exception types by replacing fragile direct struct access with a polymorphic Elixir behaviour-callback dispatch pattern.
   • Elixir
   • Phoenix LiveView
   • Ecto
   • PostgreSQL
   • Stripe
   • ExUnit
   • GitHub Actions
   • CI/CD

3. Data Analytics Intern

   Jan 2026 – May 2026

   Institutional Research Department, Gettysburg College · Gettysburg, PA

   • Maintained and updated the Institutional Research website in HTML, ensuring webpages stayed current for public access.
   • Automated class size, retention, and graduation reporting across 20+ peer institutions using Excel Pivot Tables and VLOOKUPs, cutting processing time by 30%.
   • Analyzed faculty salary data against 15 peer institutions in Excel, feeding a compensation review report to the Provost's office.
   • Excel
   • HTML
   • Data Analysis

4. AI Engineer Intern

   Jun 2025 – Aug 2025

   KPMG · Berlin, DE

   • Re-recruited based on Summer 2024 performance; bypassed the formal interview cycle to lead an AI engineering project in Agile Scrum sprints for KPMG's Global Audit division serving international teams.
   • Architected a vector-graph RAG pipeline (LangGraph, Neo4j) with CoT prompting and Instructor-structured outputs, generating 500+ judge-validated QA pairs for an audit-expert LLM, saving 50+ hours.
   • Built a multi-threaded LLM-as-a-judge framework in the OpenAI SDK to validate every generated QA pair's answer quality before fine-tuning.
   • Built a GPT-4o vision parser with PyMuPDF rendering and exponential-backoff retries for 600+ audit manual pages, eliminating 10 hrs/week of manual correction.
   • Cut inference cost ~17x by routing bulk audit-question generation to GPT-4o-mini with step-decomposed prompts and Pydantic-validated outputs.
   • Modularized preprocessing into 5 documented notebooks sharing a structural chunking algorithm for both SFT and CPT pipelines.
   • Python
   • LangGraph
   • Neo4j
   • RAG
   • OpenAI SDK
   • LLM-as-a-judge
   • Pydantic
   • Fine-tuning

5. Data Scientist Intern

   May 2024 – Aug 2024

   KPMG · Berlin, DE

   • Improved KPMG Clara 2.0 audit-risk clustering 15% in cohesion/separation by engineering a stochastic algorithm atop a rule-based baseline in Python.
   • Selected the top clustering method from 5 approaches across 15 Azure DevOps sprints, validating accuracy with certified auditors.
   • Enabled risk modeling on 500+ failed banks across two decades using Python, git, and KNN imputation, closing 30% ERP data gaps.
   • Python
   • Unsupervised Learning
   • Clustering
   • Azure DevOps
   • Git

Education

Honors & awards

• 🏆
  Earl E. Ziegler Junior Mathematics Award

  Gettysburg College · Oct 2025

  Awarded to the mathematics major with the highest average in mathematics through the junior year.

• 🏆
  Larry & Linda Oman Academic Scholarship

  Sigma Chi Foundation · Aug 2025

  A selective scholarship recognizing leadership, academic motivation, and commitment to chapter and community.

• 🏆
  Distinction Award (National Top 20)

  Australian Intermediate Mathematics Olympiad · Nov 2019

• 🏆
  High Distinction Award (National Top 20)

  Australian Intermediate Mathematics Olympiad · Nov 2018