About
JonesHong
From GPT Chat to Claude Code — a super-individual in the making.
- Systems architect
- Applied AI
- Full-stack engineer
- Blogger
Backstory
Hi, I'm JonesHong.
My interests have run wide since university. From taking hardware apart and replacing screens and batteries to flashing Android firmware, recovering data, and comparing subscription plans across regions — if it's a device problem, I'm always happy to roll up my sleeves and fix it. Later I built my own streaming setup too, from recording company meetings to long work-with-me streams.
Before graduating I was already doing frontend development as an intern. Over the seven-odd years since, I've grown from frontend presentation into database design, full-stack development, and now systems architecture. In recent years I've focused on bringing AI into real-world settings, with projects including a virtual agent for metro stations, a Q&A system over medical reports, and safety-incident detection on industrial sites.
My AI tools went from ChatGPT and Claude to Claude Code, which I now use most, and each switch let one person do a bit more. My goal is to become a "super-individual": using AI to amplify personal output and covering the full chain from system architecture through full-stack development to AI applications on my own. That doesn't mean doing everything myself, but judging what to keep in my own hands, what to hand to AI, and what to take off the shelf. Today I work with several AI agents, each handling its own part.
I can take an independent engagement from planning through delivery on my own, or join an existing team on an outsourced project. Get in touch anytime to talk about working together!
- Frontend
- Database design
- Full-stack
- Backend & AI integration
- Systems architect
- Forward deployed
- AI agent architect
Skills
- Backend
- PythonNode.jsNestJSGoRustFastAPI
- Frontend
- AngularSvelteReactTypeScriptTailwind CSS
- Databases
- PostgreSQLRedisQdrantNeo4jMongoDBMySQL
- Infrastructure
- DockerKubernetesNginxCloudflare
- 3D & Interactive
- Unreal EngineUnity
- Applied AI
- ASR / TTS / LLM integrationComputer vision integrationRAG & knowledge graphs
- AI Toolchain
- Claude CodeCodexGrokAntigravityNotebookLMStable DiffusionHermes AgentOpenClawMCPDynamic Skill LoadingSemantic Search
Quest log
Hands-on experience since university, across different settings. Three tracks running in parallel, each in date order:
Project work
-
Internship: Frontend for Fund and Logistics Systems
2017–2018, while still at university. Frontend charts and fund formula calculations for a fund system, and frontend forms for a logistics system
-
HR System 1.0 → 2.0
2017–2019. On 1.0 I built part of the frontend components; on 2.0 I led the data model and data flow design, ran the project meetings, and drove the project forward
-
AGV Robotics for a Smart Factory
In 2019, repeated trips to Xiamen, China, adding up to nearly half a year on site, working on the smart factory system design
-
Warehouse Control System (WCS)
2021, an automated warehouse project for an automation equipment maker; I built the frontend, backend, and database
-
Safety-Incident Detection on Industrial Sites
Two projects in the energy sector (2021–2024): I built the event-driven backend modules and APIs, containerized with Docker for multi-machine deployment. Colleagues set up the Kubernetes environment; I handled deployment for my own modules — editing ConfigMaps, rolling out the app, and digging into problems through k9s
-
Taoyuan Metro AI Virtual Agent "Feifei"
Launched at stations A18 and A20 in December 2022 and covered by TVBS News. I fed computer-vision results into Unreal to drive the character’s movements and the overall interaction logic. The project started on Unreal Engine 5, then a plugin pinned us to 4.27, so the whole project moved to 4.27
Interview footage from the 2022 TVBS News report on Taoyuan Metro’s 5G and AI operations -
Voice AI Pipeline
An AI interaction system for Traditional Chinese Medicine (2024–2025): an on-premise ASR (speech recognition) → LLM → TTS (speech synthesis) data flow, combined with computer vision to drive the Unity scene and interactions
-
Q&A over Medical AI Reports
User interaction for a medical AI report system (2025–2026), answering users’ questions in a Q&A format
-
Data Storage Architecture for an Enterprise AI Agent Platform
An on-premise AI agent platform (2026) spanning production, sales, HR, R&D, and finance; I owned its data storage architecture
-
Client-Facing System Design
Designed database schemas, data flows, and interaction logic against client requirements, making technical calls inside their business context. The industry now calls this role FDE (Forward Deployed Engineer)
Open source & teaching
-
Shared Libraries and Open-Source Packages
Pulled logic repeated across projects into packages published on npm and PyPI. mycena-store shipped 169 versions between 2022 and 2024; later work is mostly Python packages such as pretty-loguru, pystorex, redis-toolkit, and phonofix
-
University Lecturer: AI Image Generation Workshops
Taught twice at Feng Chia University. In April 2023, a 5-hour campus-wide workshop with nearly 80 participants from 31 departments and 3 administrative units, creating images with Stable Diffusion. In March 2024, "Stable Diffusion in Practice" for the Chinese literature department, three sessions totaling 6 hours, from prompting and text-to-image / image-to-image through LoRA training and ControlNet
The AI image generation workshop at Feng Chia University, April 2023
Side projects & tinkering
-
Flashing, Repairs, and Data Recovery
Since university. My degree covered a wide range — a bit of hardware, a bit of design, with software development as the main track. Out of curiosity I have flashed Android firmware, opened phones up to replace screens and batteries, and pulled data off an SD card that would no longer mount. For subscription services I compare what the plans look like in different regions before buying. When friends and family have phone or computer trouble, I am usually the one who takes it on, and that has not stopped since
-
Livestreaming: Starting With One Phone
Starting in 2018, I streamed on YouTube from an Android phone to record internal meetings and training videos. Curiosity took over from there, and I bought streaming gear of my own
-
"Work With Me" Streams
In 2023, as remote work became more common, I pointed the same setup at "work with me" streams — the longest ran 8 hours straight. Between company sessions and my own streams, the total runtime passed 100 hours. Later I also helped a church configure and set up its streaming rig, bringing my own equipment on site
Contact
Open to part-time and freelance projects, remote-first.