{"id":"00000000-0000-0000-0000-000000000001","contact":{"name":"Sunny Yuen","email":"sunny@yuens.me","github":"https://github.com/yuens1002","website":"https://yuens.me","linkedin":"https://linkedin.com/in/yuens-me","location":"Gaithersburg, MD","pronouns":"he/him"},"summary":"Full-stack engineer with 6+ years shipping production-grade systems, increasingly at the intersection of TypeScript, AI agents, and the Model Context Protocol. Builds and operates multi-product ecosystems as sole engineer, owning specification, architecture, implementation, QA, and operations end-to-end. Designs AI-augmented systems that are deterministic and auditable by construction: engineering judgment owns what to build and why; AI accelerates the mechanical work.","skills":[{"items":["TypeScript","JavaScript","Python","Go","HTML"],"category":"Languages"},{"items":["Next.js","React","Hono","Node.js","Tailwind CSS","shadcn/ui"],"category":"Frameworks & Libraries"},{"items":["Anthropic Claude API","Vercel AI SDK","MCP (@modelcontextprotocol/sdk)","Claude Agent SDK","OpenRouter"],"category":"AI & Agents"},{"items":["Supabase","PostgreSQL","pgvector","Prisma","Neon PostgreSQL","Railway","Vercel","GitHub Actions","Zod"],"category":"Data & Infrastructure"},{"items":["Jest","Playwright","Vitest","RTL","Cucumber"],"category":"Testing"}],"employment":[{"title":"Product Engineer","pinned":true,"bullets":["Applied systems engineering discipline to planning, implementation and testing across 1,000+ merged PRs through AI co-development tooling that checks each build against its spec before merge","Delivered Bookie, the Artisan Roast e-commerce platform and SDK, and Resume Agent within 9 months by owning every stage from specification and testing through cloud infrastructure, deployment and maintenance","Made AI features reliable across e-commerce search, agent protocols and résumé generation by replacing prompt-string tests that passed regardless of model behavior with eval-driven development using rubric scoring, LLM-as-judge and golden sets drawn from real traffic"],"company":"Self-Employed","end_date":null,"start_date":"2023-08"},{"title":"Frontend Developer","bullets":["Built a reporting and staff-directory web app for Los Angeles World Airports (LAX) with Next.js, React, TypeScript and MUI, used by all airport personnel and management layers","Integrated Microsoft Active Directory to make all LAX personnel searchable, replacing phone directories","Launched the beta and a rebranded v1 with authenticated REST APIs into live production, replacing ad-hoc forms and phone hand-offs between staff"],"company":"Wipro, Inc","end_date":"2023-08","start_date":"2023-01"},{"title":"Software Developer","bullets":["Raised unit test coverage from 50% to over 80% across the entire component library using Jest and React Testing Library","Reduced open npm security vulnerabilities to zero through automated and manual dependency updates","Refactored Coherence Design System components into type-safe pure functions, meeting Fluent UI usage requirements across Microsoft products","Audited and fixed accessibility issues until the application met WCAG AA, a hard requirement for acceptance"],"company":"DesignIt, Inc","end_date":"2022-12","start_date":"2022-04"},{"title":"Frontend Developer","bullets":["Automated Section 508 accessibility and code-quality testing for the US Census Bureau with Jest, Playwright and Cucumber, so every change was checked for compliance in CI","Refactored the Farmers.gov Drupal 8 theme into modular Sass and JavaScript components reused across the entire site","Launched a public Vue.js frontend backed by Drupal as a headless CMS for farmers applying to government-assisted programs, letting non-technical admins publish content without a developer","Kept development velocity on the agreed iteration cadence across a 100+ person SAFe Agile program by sizing and load-estimating work in planning and surfacing spec-implementation gaps in daily standups"],"company":"MetroStar Systems","end_date":"2022-03","start_date":"2020-05"},{"title":"Software Engineer","bullets":["Shipped cross-product features using React, Redux, REST and GraphQL for a real-estate intelligence app serving nationwide brokerage firms","Introduced component-level feature flags, enabling staged production rollouts for segmentation and product tiers","Built global UI components (modals, drawers) and styled components for a custom shared library with the UX team"],"company":"Remine, Inc","end_date":"2020-03","start_date":"2019-04"},{"title":"Front End Developer","bullets":["Built the DealerOn Query Language (DQL) interface so power users could query their inventory database without learning a query language","Developed client-facing interactive analytics dashboards with Vue.js and Google Charts, used across the company's nationwide dealership portfolio","Prototyped user interfaces from design mocks with stakeholders and product managers to inform product decisions and usability testing"],"company":"DealerOn, Inc","end_date":"2019-04","start_date":"2017-03"}],"education":[{"field":"Web Development","degree":"Certificate","end_date":"2016","start_date":"","institution":"Seattle University"},{"field":"","degree":"Bachelor of Arts","end_date":"1999","start_date":"","institution":"University of Washington"}],"projects":[{"url":"https://demo.artisanroast.app","name":"Artisan Roast Store","repo":"https://github.com/yuens1002/artisan-roast","role":"Sole developer and architect — designed and built the open-source store and AI-native development workflow end-to-end.","slug":"artisan-roast","tech":["Next.js","TypeScript","Prisma","Neon PostgreSQL","Stripe","NextAuth.js v5","Tailwind CSS","shadcn/ui","Jest","Playwright","Claude Agent SDK","Google Gemini","Zod","GitHub Actions","Vercel"],"cover":"https://snvyqnnmeotratupqbua.supabase.co/storage/v1/object/public/project-covers/artisan-roast.webp","impact":"Live at artisanroast.app with continuous weekly releases. Ships a free community tier, paid support plans, and add-on packages. Hosted multi-tenant solution and AI back-office MCP server in active development — a full SaaS platform built and maintained as a solo developer.","status":"active","problem":"Independent coffee shop owners lack affordable, purpose-built e-commerce tooling. Generic platforms are expensive and not tailored to specialty coffee workflows. Non-technical owners have no viable self-hosted option.","started":"2025-11","highlights":["Engineered a nightly QA system using Claude Agent SDK + Playwright that reads accessibility trees (not CSS selectors) to verify 16 acceptance criteria — auto-opens GitHub issues on regression","Designed a 'Contract-Driven' UI architecture for SaaS plan management: uses a canonical visual harness to loop through captured SDK payloads, generating 17+ state-specific UI scenarios (Trial, Expired, Pending, etc.) to prevent cross-repo drift","Implemented a 3-state asynchronous payment machine (Converting → Pending → Provisioning) with non-dismissable polling modals, generic error recovery, and ARIA-live status announcements for high-stakes billing transitions","Architected a 'Demo-Bypass' routing layer for the Plans page that intercepts platform-materialized endpoints to simulate successful checkouts in demo mode while enforcing strict contract testing for production paths","Developed a 'Provider-Driven' plan card system that eliminates slug-based conditional rendering by dispatching on SDK status (NONE/ACTIVE/TRIAL/EXPIRED/CANCELLED/INACTIVE) with inline-left Pool CTAs and management dropdowns","Built a spec drift guard (GitHub Actions) that enforces bidirectional sync between VERIFICATION.md and qa-agent.js — prevents spec/implementation divergence across the team","Optimized CI/CD reliability by engineering a custom Playwright installation strategy (using yauzl/unzip) to bypass silent extraction hangs on GitHub Actions runners, reducing build timeouts and improving deployment velocity","Built a complete coffee-native e-commerce platform (products, orders, subscriptions, admin, drag-and-drop menu builder) as a solo developer — 118 API routes, 39 Prisma models, 638 TypeScript files"],"description":"Open-source e-commerce store for specialty coffee retailers, built with an AI-native development workflow.","architecture":"Artisan Roast is an open-source e-commerce and SaaS platform for specialty coffee featuring a drag-and-drop menu builder and an AI-powered CMS for brand-aligned content generation. The system is built on Next.js 16 and React 19 using a type-safe architecture with Prisma and Neon PostgreSQL, integrating Stripe for complex subscription lifecycles. A key engineering highlight is the provider-agnostic SDK that enables build-time isolation and dead-code elimination between self-hosted stores and managed platform environments. Reliability is maintained through an automated agentic development loop and a nightly QA pipeline where a Claude Agent SDK script performs Playwright tests against the accessibility tree to verify acceptance criteria.","git_evidence":{"source":"https://github.com/yuens1002/artisan-roast","provider":"github","signature":{"alg":"ed25519","value":"DowOa7Nub9kmXFqU3hWqOq-8EHkmzrqZ6sUTVY1HeUbIZXnc3L7TeHu7A715x4jTeSn5tyccA6E_dVpI37lbBg","key_url":"https://agent.yuens.me/.well-known/oep-public-key.json","signed_at":"2026-09-07T02:04:55.487Z","fingerprint":"bZCBY6x_RnGLyQgnCY0lN7CvpiBZMvRUBo68oeisPYc"},"repo_stats":{"test_files":138,"total_files":1440,"typescript_files":1015},"verified_at":"2026-09-07T02:04:55.485Z","commit_count":1106,"contributors":2,"last_push_at":"2026-07-10","default_branch":"main","repo_created_at":"2025-11-11"}},{"url":"https://artisanroast.app","name":"Artisan Roast SaaS Platform","repo":"","role":"Sole developer and architect — designed and built the SaaS layer, provisioning pipeline, and all platform services end-to-end.","slug":"artisan-roast-platform","tech":["Next.js","TypeScript","Prisma","Neon PostgreSQL","Stripe","NextAuth.js v5","Tailwind CSS","shadcn/ui","MCP protocol","Claude API","GitHub Actions","Vercel","Zod","React","Vitest"],"cover":"https://snvyqnnmeotratupqbua.supabase.co/storage/v1/object/public/project-covers/artisan-roast-platform.webp","status":"active","problem":"The open-source store requires self-hosting expertise. The SaaS platform removes that barrier: owners get a managed instance, Stripe billing, and AI-powered tools with zero DevOps overhead.","started":"2025-11","highlights":["Sole engineer on a full commercial SaaS platform built from scratch, spanning infrastructure provisioning, Stripe billing, developer SDK, multi-state subscription lifecycle, and support tooling.","Built an ecosystem for the open-source Artisan Roast e-commerce store, providing add-on services for store owners and an SDK for independent developers to build on.","Instant rollout of new plans and services to all store instances with no redeployment required.","Automated store provisioning so a new customer signs up and gets a live store in minutes with no manual setup.","Built a billing change system that enforces legal notice periods and protects customers from surprise price increases.","Designed the subscription lifecycle to handle 8+ customer states reliably, backed by 845+ tests with full coverage on billing transitions."],"description":"Operator control plane for the Artisan Roast stack — manages app installs, subscription plans, billing, license enforcement, and provisioning automation across all tenant shop instances.","architecture":"Artisan Roast SaaS Platform is a managed hosting and subscription management service built on top of the open-source Artisan Roast store framework. The platform automates the full customer lifecycle including multi-tenant provisioning, license enforcement, and integrated billing for shop instances. The technical stack leverages Next.js for the frontend, Supabase for authentication and tenant data, and a custom SDK for validating plan and add-on configurations. The architecture prioritizes a commercial-grade multi-tenancy model that separates core open-source logic from proprietary subscription and infrastructure management.","git_evidence":{"source":"https://github.com/dev-yuen-agency/artisan-roast-platform","provider":"github","signature":{"alg":"ed25519","value":"6Bbp3n3dV-prk0_rt3w18ePZxTNzg4-jHWIbSnmFcAxgVvht5Tt81a9q6jWCZxZJ6Ybi3JuHBl83meLzy3zBCg","key_url":"https://agent.yuens.me/.well-known/oep-public-key.json","signed_at":"2026-08-06T02:59:16.851Z","fingerprint":"bZCBY6x_RnGLyQgnCY0lN7CvpiBZMvRUBo68oeisPYc"},"repo_stats":{"test_files":80,"total_files":533,"typescript_files":311},"verified_at":"2026-08-06T02:59:16.851Z","commit_count":246,"contributors":1,"last_push_at":"2026-06-16","default_branch":"main","repo_created_at":"2026-03-05"},"skipChangelog":true},{"url":"https://agent.yuens.me","name":"Resume Agent","repo":"https://github.com/yuens1002/resume-agent","role":"Sole developer and architect.","slug":"resume-agent","tech":["Node.js","TypeScript","Hono","Supabase","PostgreSQL","pgvector","Anthropic Claude","Vercel AI SDK","MCP (@modelcontextprotocol/sdk)","OpenRouter","Zod","Railway","GitHub Actions"],"impact":"Designed to be discovered and queried by employer AI systems, ATS tools, and personal LLM agents. Queryable via QR code at networking events or by any A2A-compatible system.","status":"active","problem":"AI systems are increasingly the first pass in hiring. Most candidates have no structured, queryable representation of their professional profile that AI agents can interrogate directly.","started":"2026-03","highlights":["Production-grade MCP implementation: SSE streaming transport with 30s keepalives, OAuth2 (Auth Code + PKCE) for Claude.ai integration, and dual-visibility servers (private tools vs. public recruiter tools)","Multi-stage Latency Optimization: Reduced behavioral p95 from 11.3s to 320ms via two-layer response caching (in-process + pgvector-aware), per-category maxToken tuning, and progressive disclosure for enumeration queries","A2A (Agent-to-Agent) v1.0 compliance: RFC 8615 canonical /.well-known/agent.json discovery, Ed25519 domain verification (OEP), and signed git-evidence attestation for verifiable project history","Truth contract enforced by code: inline [N] citation rules and a 18-case eval harness (deterministic rubric + LLM-as-judge) that catches fabrication, confidence inflation, and citation drops","UX-optimized Query Engine: Dual-mode interaction (conversational prose vs. structured JSON), third-person follow-up suggestions, and progressive disclosure to bound LLM costs while maintaining quality","Nightly GitHub-to-OB1 Sync: LLM-powered architecture reconciliation via semantic diffing (Gemini 3 Flash via OpenRouter) to extract project thoughts from CHANGELOGs and automated version-drift detection between package.json and logs","Observability & Eval Discipline: Automated median-of-N latency baselines, phase-timing instrumentation (LLM vs. retrieval), and majority-vote correctness evals to resolve behavioral-fixture flakes","Deterministic job-fit scoring: AI provides categorical sub-dimension judgments (skills, experience, domain), while code computes the weighted average (50/30/20) to eliminate LLM drift and score inflation"],"description":"A2A-compatible AI agent that represents a professional profile as a machine-queryable JSON API — built for employer AI systems, ATS tools, and personal LLM interfaces.","architecture":"This project provides a machine-queryable AI agent that serves as a canonical, truth-grounded professional profile for employer AI systems and personal use. Built on Node.js and Hono, the architecture utilizes the Vercel AI SDK to orchestrate reasoning tasks via Anthropic and OpenAI models while leveraging the OB1 framework for knowledge capture and storage. The system exposes a public HTTP API, an OpenAPI-compliant agent card, and dual MCP servers for both public tool-calling and private knowledge management. A key engineering focus is the Truth Contract, which enforces inline citations and prevents hallucination through a deterministic evaluation harness and strict system prompt rubrics. The project further implements the Open Employment Protocol Phase 1, using DNS fingerprints and Ed25519 keys to ensure cryptographic identity verification and self-sovereign data ownership.","git_evidence":{"source":"https://github.com/yuens1002/resume-agent","provider":"github","signature":{"alg":"ed25519","value":"OpZXa5aMZDoa40lsR0q3ud4yF727juA3HgqZbV8bEINoChTqHhvsm6bNaAfYOYluLAXeBayd2UjesFZ79GprBQ","key_url":"https://agent.yuens.me/.well-known/oep-public-key.json","signed_at":"2026-09-07T02:05:32.686Z","fingerprint":"bZCBY6x_RnGLyQgnCY0lN7CvpiBZMvRUBo68oeisPYc"},"repo_stats":{"test_files":40,"total_files":174,"typescript_files":108},"verified_at":"2026-09-07T02:05:32.686Z","commit_count":217,"contributors":2,"last_push_at":"2026-09-06","default_branch":"main","repo_created_at":"2026-03-22"}},{"url":"https://brew-guide-production.up.railway.app","name":"Brew Guide","repo":"https://github.com/yuens1002/brew-guide","role":"Sole developer — designed and built the MCP server, recommendation engine, database schema, and all tooling end-to-end.","slug":"brew-guide","tech":["TypeScript","Hono","Neon PostgreSQL","Prisma","MCP (@modelcontextprotocol/sdk)","Vitest","Railway","Node.js","Zod"],"cover":"https://snvyqnnmeotratupqbua.supabase.co/storage/v1/object/public/project-covers/brew-guide.webp","impact":"Public MCP endpoint at brew-guide-production.up.railway.app/mcp — usable by any MCP-capable tool. Submitted as an entry to the Hermes Agent Challenge 2026.","status":"active","problem":"Brew parameter guides are anecdotal and inconsistent — different sources recommend wildly different temperatures, ratios, and grind sizes with no basis in collective data.","started":"2026-05","highlights":["Engineered a deterministic recommendation engine that calculates confidence tiers (High/Medium/Low) based on community data volume rather than LLM probability.","Designed a 3-tier origin verification system (Exact, Fuzzy, Unknown) that programmatically degrades recommendation confidence to prevent 'hallucinated' consensus from imprecise data.","Architected a dual-journey API flow (Query → Log → Compare) supported by a unified snake_case data model and Neon Postgres + Prisma ORM for strict type safety.","Implemented a suite of 5 MCP tools over a public Streamable HTTP endpoint, enabling agentic coffee brewing via Claude Desktop, Cursor, and other MCP clients.","Developed a 'Delta Analysis' engine that provides structured feedback on variance across 4 dimensions (temp, time, ratio, grind) relative to method-specific defaults.","Established a rigorous CI/CD pipeline with a 56-test Vitest suite and auto-deployment to Railway, ensuring zero-regression logic for complex scoring algorithms.","Designed a 'Literature Automation' scaffold to ingest unstructured forum discussions, bridging the gap between community tribal knowledge and relational data structures."],"description":"Community-powered MCP server that answers 'how should I brew this coffee?' — synthesizes consensus brew parameters from logged experiments with zero LLM dependency.","architecture":"Brew Guide is a community-driven recommendation engine that aggregates coffee brewing experiments into deterministic weighted consensus parameters for specific origins and methods. The system is built on Node.js 24 and Hono 4, utilizing Neon Postgres with Prisma for persistence and Vitest for a rigorous test suite. A key architectural decision is the exposure of the engine via the Model Context Protocol using a streamable HTTP transport, allowing AI agents to query brewing data without relying on LLMs in the hot path. This setup ensures high-confidence recommendations categorized by data density tiers while maintaining an open-access origin policy for seamless integration with tools like Claude Desktop and Hermes.","git_evidence":{"source":"https://github.com/yuens1002/brew-guide","provider":"github","signature":{"alg":"ed25519","value":"5rr5OZ2B2lUZ6ff18QmILI33zOyurYhRZUCrZvxrvDM9hLF1-ZwJLdh9Gonm-KDPi_E9LbZva_MHdFivDGy_Cg","key_url":"https://agent.yuens.me/.well-known/oep-public-key.json","signed_at":"2026-09-07T02:05:53.948Z","fingerprint":"bZCBY6x_RnGLyQgnCY0lN7CvpiBZMvRUBo68oeisPYc"},"repo_stats":{"test_files":7,"total_files":105,"typescript_files":26},"verified_at":"2026-09-07T02:05:53.948Z","commit_count":58,"contributors":1,"last_push_at":"2026-06-07","default_branch":"main","repo_created_at":"2026-05-25"}},{"url":"https://www.yuens.me","name":"Resume Agent Web","repo":"https://github.com/yuens1002/resume-agent-web","role":"Sole author — architecture, design, and implementation","slug":"resume-agent-web","tech":["TypeScript","React","Vite","Hono","JSON-LD","llms.txt","MCP","A2A","Vitest"],"status":"active","problem":"The resume-agent backend exposes a rich JSON API but had no human-friendly or AI-discoverable frontend. Recruiters, hiring managers, and AI agents had no ergonomic way to explore a candidate's profile.","started":"2026-05","highlights":["Dual-mode architecture: conversational UI for humans + JSON-LD / llms.txt for AI agents and crawlers","Implements A2A agent discovery via .well-known/agent-card.json for automated recruiter-agent handshakes","Multi-modal 'Source Attribution' system that maps LLM citations to specific resume sections with interactive hover-states","Context-aware 'Suggested Questions' system that dynamically updates based on conversation history to reduce user friction","Streaming-first UX with optimistic UI updates and automatic scroll-anchoring for real-time LLM responses","Integrated Markdown and LaTeX rendering engine for high-fidelity display of technical resumes and academic credentials","Responsive 'Glassmorphism' design system with full keyboard accessibility and ARIA-compliant message threading","Open-source and generic — any resume-agent user can self-host by pointing two env vars at their backend"],"description":"A talkable, machine-readable face for any resume-agent backend — conversational UI for people, JSON-LD + llms.txt for AI.","architecture":"This project provides a conversational, AI-native front-end that transforms a static profile into a queryable interface for both human recruiters and autonomous agents. Built as a React and Vite single-page application, it utilizes a Hono middleware layer to serve structured machine-readable endpoints like llms.txt and JSON-LD alongside the visual UI. The architecture employs a server-side proxy pattern to securely handle sensitive API keys while decoupling the presentation layer from a separate backend data source via environment variables. By integrating real-time job-fit scoring and tailored resume generation, the system functions as a universal, open-source template for verifiable professional identity.","git_evidence":{"source":"https://github.com/yuens1002/resume-agent-web","provider":"github","signature":{"alg":"ed25519","value":"ryVrS0TwuRGbhFV-IW_i2fJMrdCB7hzANDxcXh1LoKF8rx_W-hmqLarad9gNJni8Y1Cs8KaxcmaSlfqs0YrpBA","key_url":"https://agent.yuens.me/.well-known/oep-public-key.json","signed_at":"2026-09-07T02:06:19.139Z","fingerprint":"bZCBY6x_RnGLyQgnCY0lN7CvpiBZMvRUBo68oeisPYc"},"repo_stats":{"test_files":2,"total_files":68,"typescript_files":27},"verified_at":"2026-09-07T02:06:19.138Z","commit_count":51,"contributors":1,"last_push_at":"2026-08-12","default_branch":"main","repo_created_at":"2026-05-29"}},{"name":"Open Employment Protocol (OEP)","repo":"https://github.com/yuens1002/open-employment-protocol","role":"Creator — protocol design and candidate-side reference implementation.","slug":"open-employment-protocol","tech":["A2A agent-card","MCP","Ed25519 signing","DNS-TXT verification","OpenAPI","Hono","TypeScript"],"impact":"Reframes hiring from value-extraction to value-transfer — turning a static résumé into a queryable, verifiable, self-sovereign node. Exercised in the wild before any formal launch: 66 real external queries over 62 days from interviewer, ATS, and AI-agent callers.","status":"active","problem":"Hiring has become a race to the bottom: candidates spam AI-optimized résumés, employers auto-reject with AI, and the engineers who would actually work together never speak first. Identity and trust stay locked inside walled gardens that profit from the friction.","started":"2026-03","highlights":["Candidate-side reference implementation is live and open-source (resume-agent + resume-agent-web, MIT); identity derived from a queried profile so a fork points two env vars at its own backend and becomes that person's site","Public, browsable evidence trail (/observations): each dated premise individually addressable and citable — verify-me-not-trust-me","Names its own biggest open problem — how a reconciled position becomes a governed, published standard — and turns it into a call for standards-experience collaborators (GOVERNANCE.md)","CA-style federated trust that stores nothing: DNS-TXT domain verification + Ed25519 co-signatures"],"description":"An open, federated protocol for hiring — a reconciler that poses the candidate and employer positions and settles a net outcome no single party owns.","architecture":"Open Employment Protocol acts as a neutral reconciler between sovereign candidate agents and employer-side role agents to settle hiring outcomes through a verifiable evidence graph. The architecture consists of three independent repositories—a candidate-side agent, an employer-side recruiter, and the OEP reconciler—utilizing DNS-TXT domain verification and Ed25519 co-signatures for a certificate-authority-style trust model. Engineering is centered on a stateless handshake mechanism where every claim resolves to falsifiable evidence via a dated observations endpoint, ensuring no single party owns the standard. This design prioritizes value transfer over black-box automation by maintaining an open, queryable graph that routes the matching market around centralized captors.","git_evidence":{"source":"https://github.com/yuens1002/open-employment-protocol","provider":"github","signature":{"alg":"ed25519","value":"wMei6O1Yblp7KFp7g_6RGz9WYBFxuvQIe4w4Pgt-Fvles5bfTGZwGsTp11qnQdcIldZo3OaavTe5IQcRdwDWCw","key_url":"https://agent.yuens.me/.well-known/oep-public-key.json","signed_at":"2026-09-07T02:06:21.913Z","fingerprint":"bZCBY6x_RnGLyQgnCY0lN7CvpiBZMvRUBo68oeisPYc"},"repo_stats":{"test_files":0,"total_files":11,"typescript_files":0},"verified_at":"2026-09-07T02:06:21.913Z","commit_count":6,"contributors":1,"last_push_at":"2026-06-02","default_branch":"main","repo_created_at":"2026-06-01"}},{"url":"https://railway.com/deploy/X2T0sN?referralCode=X2mmSS","name":"Bookie","repo":"https://github.com/yuens1002/bookie","role":"Author","slug":"bookie","tech":["TypeScript","MCP","Hono","Prisma","PostgreSQL","Neon","Zod","S3-compatible object storage (Railway Buckets via AWS SDK — not AWS-hosted)","Resend","Railway","OAuth 2.0","MCP (@modelcontextprotocol/sdk)","Vitest"],"status":"active","problem":"Financial software demands a dedicated UI — but the LLM already reads CSVs, sees receipt photos, and writes prose. Bookie owns what an LLM shouldn't improvise: a correct double-entry ledger, integer-cent math, deterministic categorization rules, and reproducible US Schedule C tax reports.","started":"2026-06","docsPath":"docs/ARCHITECTURE.md","urlLabel":"Deploy on Railway","highlights":["Multi-transport MCP server: stdio for local IDEs and HTTP with OAuth 2.0 + PKCE for remote mobile LLM access","Self-healing CLI setup: handles cross-platform CRLF normalization, interactive Neon org resolution, and idempotent project creation with guided error recovery","Automated 'Smoke Test' CI: verifies npm package integrity by installing into a disposable Neon environment and testing stdio responsiveness via the MCP Inspector","Cost-aware serverless architecture: replaced 60s polling with opportunistic cleanup and event-driven sweeps, preventing idle compute from exhausting Neon's 'scale-to-zero' allowance","Defensive integration testing: prevents data corruption via hostname normalization guards that block test suites from executing against production databases","Graceful degradation UX: handles mobile LLM constraints (like Claude.ai's sandboxed runtime) by falling back to structured data extraction when binary uploads are blocked","Observability-driven error handling: replaced opaque 'Unexpected Token' failures with shared diagnostic helpers that capture raw response text, exposing cross-test database concurrency issues","Double-entry ledger with integer-cent math — no floating-point rounding errors across all accounting operations"],"description":"MCP server for personal & business double-entry bookkeeping — no UI, the LLM is the interface","architecture":"Bookie is a TypeScript-based Model Context Protocol server that manages a double-entry accounting ledger through a Neon Postgres database and Prisma ORM. The system enforces a strict separation of concerns where the host LLM handles natural language processing while the server executes deterministic money math using integer minor units to avoid floating-point errors. It supports dual transports via stdio for local use and a Hono-based HTTP layer for cloud deployment, featuring a full OAuth 2.0 flow with PKCE for secure integration with the Claude.ai connector. Engineering highlights include a centralized remote database strategy using Neon branches for environment parity, automated tool documentation generation, and a graceful degradation pattern for receipt storage using S3-compatible Railway Buckets.","git_evidence":{"source":"https://github.com/yuens1002/bookie","provider":"github","signature":{"alg":"ed25519","value":"VzVJO_0hczlQGOJ49-KEaAWTshp3Uyb5umMp14Nu_h786GAMFWErDB4lIqoeoHLopBI8iwaG0owbDSuBNuxEBA","key_url":"https://agent.yuens.me/.well-known/oep-public-key.json","signed_at":"2026-09-07T02:06:45.708Z","fingerprint":"bZCBY6x_RnGLyQgnCY0lN7CvpiBZMvRUBo68oeisPYc"},"repo_stats":{"test_files":18,"total_files":105,"typescript_files":57},"verified_at":"2026-09-07T02:06:45.708Z","commit_count":62,"contributors":2,"last_push_at":"2026-09-02","default_branch":"main","repo_created_at":"2026-06-07"},"skipChangelog":false},{"url":"https://stayops-nine.vercel.app","name":"StayOps","repo":"https://github.com/yuens1002/stayops","role":"Solo engineer — architecture, schema and protocol design, backend, frontend, design system, test and eval harness, multi-agent delivery orchestration.","slug":"stayops","tech":["Next.js 16","React 19","TypeScript","Tailwind 4","shadcn/ui","Drizzle ORM","Neon Postgres","Clerk","Vercel AI SDK","Vercel AI Gateway","Vercel Blob","Vitest","Zod","Next.js","React"],"status":"active","problem":"Short-term-rental owners run their portfolio across disconnected tools — bookings originate on Airbnb and Furnished Finder, turnovers are coordinated by text, contractors submit work with no verifiable evidence, and guests ask the same questions every stay. StayOps unifies booking ingestion, turnover work orders, contractor checklists, and guest concierge behind a single agent, with three tokenized surfaces (owner, guest, contractor) and no app install for guests or contractors.","started":"2026-07","docsPath":"docs/PLAN.md","highlights":["UI design thesis: a paradigm shift that fuses chat and web into a cohesive melting pot where the same contract-validated component catalog renders both inside the conversation and as page surfaces, making the graphical layer the agent's primary output medium.","Designed a typed A2UI protocol and froze a 12-component catalog as a contract, with Zod validation at the dispatch boundary and a safe fallback mechanism to prevent malformed model outputs from crashing the stream.","Implemented a multi-agent orchestration pipeline for project delivery, utilizing parallel implement-and-adversarially-verify loops where every acceptance criteria verdict is backed by a verified test run or in-browser evidence.","Built a high-integrity two-way iCal sync engine: UID-keyed upsert with hash short-circuiting, tokenized per-unit export feeds, and turnover side-effects that auto-create on checkout and auto-cancel unstarted tasks.","Engineered a 'changes via agent' lock chip UX pattern, ensuring the conversation is the sole mutation path and dashboard tiles act as agent launchers that stage requests rather than mutating inline.","Shipped an eval harness with a golden owner-scenario suite that validates confirm-gating invariants in CI without model calls, supported by 150+ tests across contract, service, ingestion, and agent-pipe layers.","Enforced server-side trust invariants: a submit gate refusing work orders with missing required artifacts, immutable booking dates for synced records, and a mandatory confirmation handshake for all destructive agent actions."],"description":"Chat-first property-operations app for short-term-rental owners, where the agent is the only way state changes.","architecture":"StayOps is a personal property management platform that synchronizes short-term rental bookings via two-way iCal feeds and enriches guest data through Gmail API ingestion. The system is built on Next.js 16 and Neon Postgres using Drizzle ORM, featuring a chat-first SPA architecture powered by the A2UI protocol and a custom Tailwind 4 component catalog. Engineering patterns prioritize a prescribed, checklist-based workflow for contractors and a conversational guest concierge integrated via the Vercel AI SDK and Gateway for model-agnostic reasoning. The architecture deliberately avoids payment processing to focus on operational validation, maintaining a strict audit trail of tool invocations and persisting state within a single-tenant environment.","git_evidence":{"source":"https://github.com/yuens1002/stayops","provider":"github","signature":{"alg":"ed25519","value":"Pxo9qn1GH-gdNidXqQIuG4KwYdCKfja3oXZzOcXKZmpd72tUKNJxxIGm9emQpfS4CQgghXI-pzjlbS_iJDLbCQ","key_url":"https://agent.yuens.me/.well-known/oep-public-key.json","signed_at":"2026-09-07T02:07:08.457Z","fingerprint":"bZCBY6x_RnGLyQgnCY0lN7CvpiBZMvRUBo68oeisPYc"},"repo_stats":{"test_files":24,"total_files":268,"typescript_files":191},"verified_at":"2026-09-07T02:07:08.456Z","commit_count":50,"contributors":1,"last_push_at":"2026-08-11","default_branch":"main","repo_created_at":"2026-07-03"},"featureDocsGlobs":["docs/features"]},{"name":"OpenClaw Control Plane","repo":"https://github.com/yuens1002/openclaw-control-plane","role":"Sole developer and architect — designed and built the provisioning shell, contracts, and deployment discipline end-to-end.","slug":"openclaw-control-plane","tech":["TypeScript","Node.js","Railway","Docker","GitHub Actions"],"status":"active","problem":"Deploying and re-deploying an AI agent runtime for each new client normally means manual wizard configuration and untracked drift between instances. Nothing verifies that a live deployment still matches what was intended.","started":"2026-08","highlights":["Built an idempotent setup-profile applier that drives a live OpenClaw instance's setup API from a declarative profile, replacing manual wizard configuration for every client onboarding","Engineered a build-time wrapper patch for scoped state export, cutting a full-instance backup from 541MB to roughly 7MB while preserving exact restore compatibility","Implemented HMAC-SHA256 GitHub App webhook verification ahead of the proxy's body parser, with a JSON-configured per-repo allowlist gating what can dispatch into an isolated agent turn","Added live Railway proof checks that verify a deployment's source, runtime settings, and health endpoints against the pinned template, catching infrastructure drift automatically","Designed a workflow-neutral public core with zero client-specific assumptions baked in, letting private profile repos attach automations after onboarding without touching the shared baseline"],"description":"A governed, auditable provisioning shell for AI agent runtimes on Railway, separating reusable operating discipline from client-specific automation.","architecture":"OpenClaw Control Plane is a TypeScript monorepo that provisions and configures OpenClaw agent runtime instances without becoming the runtime itself. An idempotent setup-profile applier reads a declarative client profile and drives a live instance's own setup API automatically, replacing manual configuration for every onboarding. Build-time wrapper patches add scoped state export/import and verified GitHub App webhook handling ahead of the proxy's body parser, and live Railway 'proof' checks continuously verify a deployment's source, settings, and health against what was intended. The public core stays workflow-neutral, with private agency profile repos attaching client-specific connectors and credentials after onboarding.","featureDocsGlobs":["docs/adr"]}],"availability":{"remote":true,"status":"actively-looking","seeking":true,"preferred_roles":["Frontend Engineer","Full Stack Engineer","Software Engineer"],"preferred_locations":[]},"updated_at":"2026-10-22T02:05:11.161+00:00","tagline":"Full systems. Every layer. One engineer.","publications":[{"date":"2026-08-14","slug":"beyond-the-chatbot-chief-of-staff-control-plane","tags":["ai","agents","architecture","opensource"],"title":"Beyond the Chatbot: Why the Future of Workplace AI Needs a \"Chief of Staff\" Control Plane","platform":"Dev.to","grounded_in":"yuens1002/openclaw-control-plane (public repo) — architecture, M1 status, Railway proof checks","canonical_url":"https://dev.to/yuens1002/beyond-the-chatbot-why-the-future-of-workplace-ai-needs-a-chief-of-staff-control-plane-2jhm"},{"date":"2026-09-14","slug":"hiring-is-a-black-box-on-both-ends","tags":["ai","hiring","career","jobsearch"],"title":"Hiring Is a Black Box on Both Ends","platform":"Dev.to","grounded_in":"knowledge-base wiki/concepts/hiring_01.md (findings 2026-09-08 → 2026-09-14: Greenhouse Real Talent, GoodTime 2026 TA survey, Kickresume field experiment, Cui/Dias/Ye, Takano) + first-party job-hunt pipeline metrics 2026-09-14; draft cot-memory outreach/drafts/2026-09-14-hiring-black-box.md; decision record state/decisions/2026-09-14-hiring-black-box-outreach.yaml","canonical_url":"https://dev.to/yuens1002/hiring-is-a-black-box-on-both-ends-53dc"}],"verification_status":{"domain":"verified","git_evidence":"partial","peer_attestations":"pending","employer_cosignature":"pending_ecosystem","note":"Domain ownership verified via OEP Phase 1 (Ed25519 + DNS TXT fingerprint). Git evidence available for 8 of 9 project(s). Employment verification via employer co-signature is pending ecosystem adoption. Peer attestations via public LinkedIn posts are the current best available signal — the candidate welcomes LinkedIn recommendations from former colleagues."}}