I help people do more meaningful work by removing unnecessary friction with AI.
I turn ambiguous real-world workflows into working, testable systems—from discovery and requirements through configuration, integration, validation, user enablement, and handoff.
My strongest work sits between people, operations, and technology. I learn how the work actually happens, translate that understanding into requirements and business rules, coordinate implementation, verify expected versus actual behavior, and leave behind documentation another person can use.
I use AI to expand my speed and technical range while remaining accountable for objectives, scope, requirements, architecture, privacy, validation, communication, consequential decisions, and final approval.
Los Angeles, California · Pacific Time
Primary focus: AI enablement, technical implementation, workflow automation, and customer-facing technical solutions roles where real operational needs must become reliable, usable systems.
Discover → Structure → Configure & Integrate → Validate → Enable → Document → Improve
01 · Can AI Yet? · AI capability evaluation
A working evaluation system for testing whether AI agents and tool integrations can perform real business workflows reliably. I define bounded contracts, build evals, run expected-vs-actual checks, preserve machine receipts and visual evidence, correct failures through independent review, and keep unproven capabilities explicitly unproven — using proof states such as PROVEN, DRY-CERTIFIED, CONFIGURED_NOT_LIVE_PROVEN, BLOCKED, and NOT_PROVEN.
The live product is an early alpha under active development; its interface, catalog, and evidence will continue to change. Current integration evidence includes bounded HubSpot sandbox work, not an enterprise rollout or proof of production customer deployment. The public case study is the canonical recruiter-facing evidence summary.
Read the visual summary → · Explore the public case study → · Open the live alpha →
02 · Paid independent client implementation · Client delivery and handoff
Took a real client need from discovery into a client-operable production system: public Next.js experience, client-owned GitHub → Vercel path, consultation lead intake into Google Sheets, duplicate-submission protection, Safari/Chrome and mobile validation, seller-guide enablement surfaces, documentation, and ongoing iteration. Client identity and private details stay protected.
03 · Hack for LA — TDM Calculator · Volunteer Front-End Contributor
Contributed as a volunteer—not an employee—to Hack for LA's TDM Calculator, a React/Vite civic application. The work shows scoped implementation in an existing team codebase, automated validation, and changes carried through public peer review.
PR #3531 standardized external-link behavior across the application, added test coverage, addressed requested changes, and was merged after approval. PR #3576 adds and tests a submissions empty state; it is open and under review, not merged.
Review the merged contribution → · See the current work under review →
04 · Autonomous Trading Systems Lab · Requirements, validation, and human gates
A public-safe case study showing how I direct AI-assisted engineering: turning an ambiguous, high-consequence objective into bounded requirements, deterministic authority, sequenced integrations, observable validation, independent review, and human approval before consequential action. It is research documentation, not a production financial system.
The public case study was reconciled with current project status on 2026-10-04. Phase 5 remains in progress and deliberately unmerged; the protected private main branch remains fail-closed. It does not claim current runtime readiness, an ALLOW path, profitable trading, positive expectancy, or live brokerage execution.
Open the visual case study → · Read the full case study → · Review the repository →
05 · Cousin Radio · Shipped product
Led a live family music product from user observation through requirements, AI-assisted implementation, troubleshooting, QA, deployment, and continued iteration.
Visit the live product → · Review the employer evidence → · Explore the public showcase →
| Capability | What that means in practice | Evidence |
|---|---|---|
| Workflow discovery and scoping | Understand the user, current process, friction, dependencies, constraints, and definition of success. | Paid client implementation · Cousin Radio |
| Requirements and system design | Convert ambiguity into business rules, acceptance criteria, owners, boundaries, and an executable implementation path. | Autonomous Trading Systems Lab · Paid client implementation · Can AI Yet? |
| Configuration and integration | Connect tools, interfaces, environments, data flows, and deployment workflows in a deliberate sequence. | Paid client implementation · Hack for LA — merged PR #3531 · Can AI Yet? |
| Validation and diagnosis | Compare expected versus actual behavior using tests, logs, screenshots, browser evidence, CI, and repeatable checks. | Can AI Yet? · Autonomous Trading Systems Lab validation gates · Hack for LA — merged PR #3531 |
| User enablement and handoff | Explain the system clearly, conduct walkthroughs, document ownership, separate completed work from open dependencies, and leave a usable next step. | Paid client implementation · Career Development OS |
| AI workflow evaluation | Design evals, preserve machine evidence, run correction/review loops, and keep unproven capabilities labeled unproven. | Can AI Yet? · Implementation + AI systems |
| AI-assisted execution | Coordinate AI tools for research, implementation, testing, and documentation while keeping human accountability and approval explicit. | Implementation + AI systems · Model and tool attribution |
- Start with the person, the process, and the real operational friction.
- Define success, constraints, ownership, and what must remain human-controlled.
- Translate ambiguity into requirements, business rules, acceptance criteria, and tracked work.
- Sequence implementation around dependencies, risk, and the fastest path to trustworthy evidence.
- Use AI and technical tools to extend my speed and range without outsourcing judgment.
- Validate expected versus actual behavior before presenting work as complete.
- Enable the user through clear communication, walkthroughs, documentation, and handoff.
- Turn repeated learning into reusable workflows, playbooks, and system improvements.
flowchart LR
discover["Discover workflow"] --> requirements["Structure requirements"]
requirements --> configure["Configure or integrate"]
configure --> validate["Validate expected vs. actual"]
validate --> humanGate{"Consequential action?"}
humanGate -->|"Yes"| humanReview["Human review"]
humanGate -->|"No"| enable["Enable the user"]
humanReview --> enable
enable --> handoff["Document the handoff"]
handoff --> improve["Improve the next implementation"]
AI accelerates my research, implementation, testing, and documentation. I remain responsible for defining the objective, setting boundaries, deciding requirements, reviewing outputs, validating behavior, protecting privacy, communicating status, and approving the final result.
I do not present AI-generated output as proof by itself. A claim becomes credible when it is supported by working software, tests, logs, screenshots, deployment evidence, documented decisions, or a reproducible workflow.
Read how I use AI and technical tools →
- Career Development Operating System — a public-safe view of a private workflow system for research, prioritization, evidence routing, agent coordination, durable handoffs, and human decision gates.
JavaScript · React · Next.js · HTML/CSS · GitHub · Vercel · REST API and JSON concepts · Model Context Protocol · Playwright · browser developer tools · Python and SQL fundamentals · workflow mapping · forms and validation logic · expected-versus-actual QA · CI and deployment awareness · Markdown documentation · Cursor · ChatGPT · Claude Code · Gemini
- 7,000+ customer-facing interactions with a 4.9-star service record
- Current experience in high-volume customer operations
- Approximately ten years in safety-critical film and television production
- Paid independent client implementation and handoff experience
- Meta Front-End Developer, Google Cybersecurity, and Google AI Essentials certificates
- Continuing hands-on development in AI workflows, technical implementation, web delivery, Python, and human-centered systems
My background taught me to stay calm around complexity, communicate clearly across different kinds of people, adapt quickly when conditions change, and remain accountable for the final result.
LinkedIn: chien-escalera-duong
Email: heyitschien@gmail.com