YANGLIUENGINEER
► PORTFOLIO v1.2 · 2026
It doesn't matter if you call me prompt/context/harness engineer or just AI Engineer in general, I always get the job done. I build anything AI that ship work, full-stack systems that scale, and growth engines that turn ad spend into pipeline. Get in touch if you need me to build something Bold & Complex.
// handleyang-liu.dev
// roleAI ENG · FULL-STACK
// originShandong, CN
// years4.3y shipping
// locBrisbane, AU
// rate$80 / hr
// status● Available Now
[ PORTRAIT // YL-001 ]
01 // ABOUT THE OPERATORWHO_IS_THIS.MD
Engineer who treats shipping as the only metric that matters.
I work where AI, product, and growth collide — building agentic systems that don't just demo well, they pay rent.
Past four years I've shipped products across web, mobile, and infra from real time chat fleets at scale to agent graphs that triage support tickets, qualify leads, and draft ad copy that beats hand written control.
I write production code in TypeScript, JavaScript, and More. I run Google Ads campaigns with the same rigor I push commits: hypothesis, instrument, ship, measure, kill.
Not interested in vibes-only AI. Interested in agents that close tickets, campaigns that close deals, and code that closes the loop.
64
projects shipped
3.2M
agent runs served
3.7×
avg ROAS uplift
99.94
% uptime sla
02 // STACK · DOMAINS · TOOLS OF THE TRADECAPABILITIES.YML
01 / 08
■AI Agents & Workflows
02 / 08
■Web Engineering
03 / 08
■Mobile / App Dev
04 / 08
■Growth / Marketing
05 / 08
■Infra / DevOps
06 / 08
■Data & RAG
07 / 08
■Design Systems
08 / 08
■Backend Services
PROFICIENCY
MAP_v2.json
MAP_v2.json
TypeScript / React96.0
Next.js 16 / Edge94.0
LLM agents (LangGraph)92.0
Python / FastAPI88.0
Google Ads / Growth90.0
Postgres / pgvector85.0
Rust72.0
React Native83.0
03 // WORK · INDEX 01–09./projects/*.md
Things I built
that made it
to production.
Nine selected projects, each one shipped, measured, and either still running or sunset with intent. Sorted by recency, tagged by domain. Click any to read the post-mortem.
A
01 / 09AI AGENT
Orion — Multi-agent Support OS
CASE STUDY→
B
02 / 09FULL STACK
Beacon — Real-time Ops Console
CASE STUDY→
C
03 / 09GROWTH
Helix — Ads Bidding Agent
CASE STUDY→
D
04 / 09MOBILE
Drift — Travel Companion App
CASE STUDY→
E
05 / 09AI AGENT
Forge — RAG Doc Pipeline
CASE STUDY→
F
06 / 09FULL STACK
Pulse — Realtime Analytics Cloud
CASE STUDY→
G
07 / 09GROWTH
Loop — Ad Creative Generator
CASE STUDY→
H
08 / 09AI AGENT
Atlas — Code-review Bot
CASE STUDY→
I
09 / 09INFRA
Switchboard — Edge Function Mesh
CASE STUDY→
~/orion/src/agent.ts
// MANIFESTOPROD_READY
Most "AI demos" leak in production. Orion runs 14k tickets/day for a payments client — with a 87% deflection rate, $0.011 per resolve, and zero PII incidents in 11 months.
- Model-routing graph picks cheapest model that still passes eval gates
- Retrieval is hybrid — pgvector + BM25 + recency boost
- Every action is reversible, every reversal is audited
- Replay any session in <800ms from cold storage
04 // TIMELINE · 2023 → NOWHISTORY.LOG
Four years.
One throughline.
2024→ PRESENT
Founder · AI Engineer · Full-stack Dev
@ Workshown
Shipping AI agentic products for fintech and SaaS. Implemented AI Harness infrastructure for AI SEO/SEM analysis agent. Built Ecommerce for Nasdaq listed company
2024→ 2026
Full-stack Dev · AI Engineer
@ My Smart Living Inc.
Built the LLM-routing layer that backs My Smart Living's product copilot. Cut inference cost 41% while raising eval pass-rate from 71→89%. Led a team across infra & RAG.
2024→ 2026
Web Developer
@ Memory Marketing
Built dozens of websites for high-profile and local professional clients. Tech stack includes Next.js, Postgres, Stripe Connect, Vercel and Cloudflare suites.
2023→ 2023
R&D
@ R̶e̶d̶a̶c̶t̶e̶d̶
Developed micro-controller hardware for IOT applications