I build user-facing products with React and TypeScript, then carry them through data design, testing, documentation, and release. I focus on dependable local-first software where security boundaries, evidence, and failure states matter.
I’m based in Chicago and open to software engineering roles across frontend, full-stack product development, and desktop applications.
Portfolio · Résumé · LinkedIn · Portfolio source
-
DrawScope · v0.6.5 — React/Tauri research workbench for auditable lottery archives and honest historical pattern testing. Its reproducible SQLite dataset contains 41,598 deduplicated draws across six games, and its methodology uses leakage-resistant walk-forward trials instead of prediction claims. Release · Methodology
-
GameVault · v0.3.5 — Portable React/Tauri game library for user-owned files. Its review-gated archive pipeline blocks traversal, unsafe Windows paths, links, reparse entries, and decompression abuse before extraction. Release · Safety model
-
Day-Trading Teacher · v0.36.0 — Thirteen-lesson local-first environment for planning, historical replay, paper practice, journaling, and review. Rust provides deterministic decimal calculations; the product intentionally excludes brokerage access, live signals, and order execution. Release · Project tour
-
Research Studio · source-free case study — Guarded Electron/React workflow for bilingual catalog enrichment, validated against 31,521 series and 2,242,170 episodes without changing the source database hash. Model output stays untrusted until schema, evidence, and human-review gates pass. Evidence · Threat model
- Reader — Local-first Windows reading-library alpha for EPUB, PDF, text, and authorized audio.
- Media Scout — Permission-scoped Manifest V3 extension for media discovery and authorized downloads.
- SiteWipe — Safety-engineering prerelease for reviewed, target-scoped browser-data cleanup; its release gates remain open.
These repositories are available for code review, but they are not supported public binary or extension-store releases. The portfolio keeps them visually separate from the shipped projects and case study above.
- Start with the user workflow, the decision the product supports, and the failure state it must explain.
- Keep filesystem, database, process, browser, and model authority behind explicit boundaries.
- Validate imported data with schemas, provenance, deterministic rules, and human review.
- Test across UI, domain logic, persistence, security boundaries, and release paths.
- Publish evidence with product claims: documentation, checks, checksums, SBOMs, and provenance where applicable.
AI agents assist with research, implementation, test generation, and iteration across these projects. I remain responsible for product direction, architecture, technical review, verification, safety and licensing boundaries, data-source decisions, published claims, and final release approval. Agent output is treated as untrusted until it passes repository checks and human review.
Core stack: React · TypeScript · JavaScript · Rust · Tauri · Electron · Node.js · Python · SQLite · Vitest · Playwright · pytest · GitHub Actions · CodeQL
The best public contact is LinkedIn. Project bugs and proposals belong in the relevant repository; suspected vulnerabilities should use private vulnerability reporting.