Autonomous multi-model trading system that screens the S&P 500 daily, executes trades, and retrains its own reinforcement learning agent on live results.

• Stack: Python, PyTorch, Stable-Baselines3, Claude API (Haiku + Sonnet), Alpaca API, PostgreSQL, APScheduler
• AI Pipeline: Three-tier Claude pipeline — Haiku for bulk screening of 477 S&P 500 stocks → Sonnet for deep analysis → weighted composite scoring above a 7.5 threshold
• RL Agent: PPO agent (Stable-Baselines3 + PyTorch, Apple Silicon MPS) trained on 3–5 years of S&P 500 data — 54.67% win rate and positive Sharpe ratio on held-out backtest data
• Risk Management: Two-layer system with server-side GTC stop-loss orders on Alpaca and a ratchet stop that locks in profit automatically at +5%, +10%, and +15%
• Automation: 6 APScheduler daily jobs running unattended — research, trade execution, and intraday monitoring — with full PostgreSQL trade journaling feeding a weekly RL retrain loop

Distributed system health monitor with async concurrent checks, intelligent alerting, and a real-time React dashboard.

• Stack: Java 21, Spring Boot 3, CompletableFuture, AWS SNS, React, PostgreSQL
• Concurrency: Parallel health checks via CompletableFuture — non-blocking, timeout-aware, fault-isolated per service
• Alerting: Multi-channel AWS SNS notifications triggered on threshold breaches with configurable severity levels
• Dashboard: React frontend with live status indicators, historical uptime graphs, and incident timeline

End-to-end data pipeline that ingests trending stories from Hacker News and NewsAPI, transforms and ranks them with dbt, orchestrates hourly runs with Airflow, and serves everything through a live Next.js dashboard.

• Stack: Python 3.13, PostgreSQL, dbt, Apache Airflow, FastAPI, Next.js 14, Tailwind CSS, SWR
• Pipeline: Automated hourly ingestion with idempotent upserts, staging/marts transformation layers, 6 passing data quality tests
• API: FastAPI REST API with 3 endpoints — GET /trending, GET /hn/stories, GET /news/articles — with auto-generated Swagger docs
• Frontend: Responsive Next.js dashboard with auto-refresh, proportional volume bars, loading and error states per component

Full-stack app with semantic search and context-aware responses using RAG pipelines.

• Stack: React, FastAPI, OpenAI API, Pinecone, Supabase, PostgreSQL
• Technical: OpenAI text-embedding-3-small, Pinecone vector database, Google/GitHub OAuth, RAG pipeline with GPT-4o-mini
• Architecture: Notes → Supabase → OpenAI embeddings → Pinecone → Semantic search & AI-synthesized answers
• Innovation: Finds notes by meaning not keywords — search "machine learning projects" finds notes about "neural networks"

Next.js 14 finance application with AI-powered transaction categorization and real-time analytics.

• Stack: Next.js 14, TypeScript, Prisma, PostgreSQL, OpenAI GPT-4, Tailwind CSS, Recharts
• Features: AI transaction categorization, real-time budget tracking, interactive Recharts visualizations
• Quality: 96.7% TypeScript coverage, Prisma ORM for type-safe database queries

Quality Technician @ TEQ – A Toppan Company (March 2026 - Present) Precision dimensional inspection, SPC data logging, ISO 13485 non-conformance documentation, and production floor advisory in a regulated manufacturing environment

Autocare Technician @ Walmart (June 2025 - March 2026) Vehicle maintenance, diagnostics, safety inspections, and customer service

Generative AI Engineer (Internship) @ Reality AI (Dec 2024 - Apr 2025) Built Marvel AI (Teaching Assistant) and Sky AI (Career Coach) using LangChain, LlamaIndex, RAG workflows, OpenAI embeddings, and Redis Vector Search. Deployed on Google Cloud Run with LangSmith optimization.

Full-Stack Developer (Internship) @ Connectify (Oct 2024 - Dec 2024) Developed scalable web applications with Django/React/Vue, built RESTful APIs, implemented CI/CD pipelines with GitHub Actions, and collaborated in agile teams for production-ready solutions.

Software Developer (Full-Stack) (Internship) @ Radical AI (Apr 2024 - Aug 2024) Built AI career companion platform with React/Python, integrated third-party APIs, performed unit/integration testing, and maintained technical documentation.

Manufacturing Process Leader — Training Department @ Stryker Medical (Dec 2019 - Jun 2023) Led FDA/ISO 13485 compliant training programs, managed Document Change Orders (DCOs), audited production floors, and mentored teams in regulated medical device manufacturing.

Infantry Rifleman @ United States Marine Corps (Aug 2013 - Dec 2016) Ground combat operations, reconnaissance missions, weapons systems operation, and team leadership in high-stakes environments.

BS Computer Science — Software Engineering Southern New Hampshire University (Graduated: November 2025)

• Honor Roll: Spring 2023, Fall 2023, Spring 2024
• Dean's List: Fall 2024, Spring 2025, Fall 2025

Relevant Coursework: Software Development Lifecycle, Software Security, Reverse Engineering, Data Structures & Algorithms, Full Stack Development, Mobile Architecture, System Analysis & Design, Database Environments

• 🌐 Portfolio: nwyrwas.github.io
• 💼 LinkedIn: Nicholas Wyrwas
• 📧 Email: nick.wyrwas@outlook.com

Marine Corps veteran • CS graduate • Always building • Always learning