Hi, my name is
I'm a software engineer who builds platforms that people actually use. I've helped uninsured families find free healthcare, scaled learning apps to thousands of users, and built intelligence systems for Fortune 500 companies. I write code that solves problems, not resumes.
Hello! I'm Pawel, a software engineer who's been building things for the web since I was 12. I started with Adobe Flash games on Kongregate that reached over 50,000 gameplays, and that early experience sparked a passion that's shaped my entire career.
Today, I've built AI-powered health platforms at a social impact startup, developed machine learning pipelines at Adobe, and engineered cybersecurity infrastructure for the Department of Defense. I specialize in full-stack development with a focus on AI integration, vector search, and data engineering.
Hello! I'm Pawel, a software engineer who's been building things for the web since I was 12. Back then, I was creating Adobe Flash games and text-based RPG games, publishing them on Kongregate where they collectively reached over 50,000 gameplays. That early experience of building something people actually used sparked a passion that's shaped my entire career.
Fast forward to today, and I've built AI-powered health platforms at a social impact startup, developed machine learning pipelines at Adobe, and engineered cybersecurity infrastructure for the Department of Defense. My work focuses on creating systems that are both technically sophisticated and genuinely useful.
Most recently, I've been extracting and parsing terabytes of insurer data to estimate healthcare costs, transforming unstructured medical billing data into structured formats that power an AI copilot. I specialize in full-stack development with a strong focus on AI integration, vector search, and data engineering, tackling complex problems like semantic search across 50,000+ services and real-time ML inference.
Here are some technologies I've been working with recently:
December 2025 - March 2026
Featured Project
An AI-powered platform helping uninsured users find affordable healthcare. Features an AI Copilot with semantic search across 50K+ services, adopted by shelters and caseworkers across D.C.
An AI-powered healthcare discovery platform helping uninsured Americans find affordable care across the DMV area. Combines a multi-turn AI Copilot with dual-collection semantic vector search (50K+ provider services), a 6-agent parallel data extraction pipeline, and programmatic SEO. Adopted by caseworkers and shelters across Washington D.C. with 100K+ Google impressions in 3 weeks.
Featured Project
An AI-powered platform helping companies calculate ROI for workflow automations. Uses RAG and a vector database of 500,000+ automation combinations to match use cases to cost-effective workflows. Scaled to 50+ beta users across 3 agencies.
An AI-powered automation ROI intelligence platform that helps companies calculate return on investment for workflow automations. Uses an advanced Retrieval-Augmented Generation (RAG) system and a vector database of 500,000+ automation combinations to match use cases to the most cost-effective workflows. Increased retrieval accuracy by 50% and scaled to 50+ consultants evaluating platform.
Featured Project
A gamified AI-powered platform for CompTIA Security+ exam prep. Built as an offline-first PWA with personalized recommendations and AI explanations. Scaled to 1,000+ monthly active users with 54% increase in engagement.
A gamified, AI-powered learning platform for CompTIA Security+ certification prep. Built as an offline-first progressive web app (PWA), SecuSpark provides personalized question recommendations and AI explanations running directly in the browser. Scaled to 1,000+ monthly active users with 54% increase in engagement rates through adaptive difficulty and RPG-style progression.
Featured Project
AI middleware for government SharePoint — routes natural language queries to structured data and semantic document search. Features NL-to-SQL, RAG over 25K+ document chunks, and a 10-panel no-code Agent Builder.
An AI-powered middleware engine for government SharePoint environments that routes natural language queries to structured data, semantic document search, and deep analysis tools. Features a 4-tool AI engine with NL-to-SQL translation, RAG over 25,566 vector-embedded document chunks with sub-100ms HNSW lookups, and a 10-panel Agent Builder for non-technical admins. Built as a monorepo with an Art Deco landing page and Fluent UI v9 agent dashboard.
Featured Project
Enterprise CRM for medical field sales — territory management over 9M+ NPI records, route optimization, Stripe billing, Salesforce sync, and offline-first PWA. 105K+ lines built in 22 days.
A full-stack, enterprise-grade CRM platform for medical field sales reps and managers. Features territory management over 9M+ NPI provider records with PostGIS spatial queries, route optimization, real-time activity tracking across 7 activity types, Stripe subscription billing, Salesforce bidirectional sync, and offline-first PWA support. 105,800+ lines of TypeScript built solo in 22 days.
Featured Project
AI-powered Medicaid filing automation across 25 U.S. states. Playwright-based agent navigates real government portals with ARIA-only automation. 20-language support and PII isolation architecture.
An AI-powered Medicaid enrollment platform that guides users through multilingual intake wizards and autonomously submits applications on real government portals across 25 U.S. states using a Playwright-based filing agent. Features ARIA-only portal automation (no computer vision), PII isolation architecture where the LLM never sees sensitive data, 20-language support including RTL, and a streaming voice assistant.
AI-powered tool to query U.S. hospital pricing data in plain English. Converts Medicare dataset into a conversational API with geospatial search, quality ratings, and sub-300ms query performance.
Media digitization and logistics platform for converting VHS, tapes, and DVDs to MP4. Automates scheduling, payment processing, and shipping workflows with Stripe webhooks and admin dashboards.
Production-ready Model Context Protocol (MCP) server for real-time NPI registry queries. Features multi-tier caching, fuzzy matching with RapidFuzz, and sub-500ms live calls for AI agent integration.
High-performance object detection system for live video streams targeting 60+ FPS. Custom-trained YOLOv8 achieving 95% precision with optimized pipeline and ONNX export for deployment flexibility.
ML-driven trading strategy using Random Forest on technical indicators. End-to-end pipeline with automated data ingest, feature engineering, hyperparameter tuning, and realistic backtrader simulations.
Analyzed CalWORKs eligibility and economic disparities across San Francisco using Census PUMS microdata. Config-driven pipeline with insightful visualizations identifying affordability gaps and assistance needs.
I'm currently looking for new opportunities and interesting projects to work on. Whether you have a question, want to collaborate, or just want to say hi, my inbox is always open. I'll do my best to get back to you!
pawelsloboda5@gmail.comLike what you see? This portfolio is built with Astro + React and is open source for you to fork, clone, and customize.
For a mostly static portfolio site like this, Astro's island architecture is perfect. Only the interactive bits load JavaScript, you sprinkle client:* where you want interactivity, and everything else is pure HTML. Next.js is great for heavy backend and interactive apps, but for content sites? Astro loads way faster.
Scan or Click to view my resume