resume

gainesville, fl · gpa 3.80 / 4.00

• coursework: data structures & algorithms, computer organization, discrete mathematics

San Diego, California

• Shipped AI-powered event-creation tooling and full-stack features on an engineering team of 10, owning each feature from spec through production deploy
• Built attendee matchmaking end to end (CSV ingestion and validation, vector-similarity matching, automated email delivery), cutting a manual 2-hour workflow to under 5 minutes
• Designed the Campaign feature’s lead-generation workflow, including a lead filter, and built its Django de-duplication and scoping logic
• Fixed 5+ production bugs in the campaign and email workflows within 24 hours each, adding edge-case tests to prevent regressions

Remote

• Co-developed a diagnostic framework that separates perception from reasoning failures in audio-language models on Google’s Massive Sound Embedding Benchmark (MSEB)
• Owned the measurement and statistics track on a 4-person team, building an evaluation framework with 8 task adapters, 10 metrics, a blind LLM as judge with cost accounting, and 750+ passing tests
• Implemented paired-bootstrap CIs, composite scores, and the Reasoning Robustness Index, and a simulation-based power analysis showing the preregistered design couldn’t reach 80% power even at 800 items per cell

Python, PyTorch, vLLM, LoRA, ALFWorld

• Built an open-source tool for testing AI agents that learn from experience, comparing writing lessons to memory against fine-tuning the model on its own successes
• Ran a controlled study on Qwen3-4B over 200 ALFWorld household tasks, written lessons raised success from 45% to 85%; fine-tuning beat it with 2× shorter prompts
• Measured how wrong self-grades affect learning, mislabeling 20% of outcomes changed success by 43 points

Python, Swift, FastAPI, PostgreSQL, Redis/Celery, React.js

• Built an iOS 17 application and 60+ endpoint FastAPI backend for peptide protocol management, dose logging, progress visualization, and LLM-generated weekly summaries
• Architected production infrastructure with JWT auth, iOS Keychain and Fernet encryption, async SQLAlchemy/PostgreSQL, background jobs, notifications, and Apple Health integrations
• Launched on the App Store and grew to 100+ users generating 1,000+ check-ins per month, converting 5% to premium membership of the product

Python, TypeScript, Node.js, Vercel

• Built a full-stack habit-tracking web app at SwampHacks that gamifies financial and health tracking through AI-generated progression levels
• Integrated the Gemini API to adjust progression tiers from behavioral data and the Capital One API for real-time transaction tracking against user spending limits