Frontend
Interfaces people actually enjoy — responsive, accessible, fast, and built on a design system instead of one-off hacks.
Open to building something ambitious
Davy Chen
FullStack Developer · Based in China
I build software end to end — interfaces, services, pipelines, deployment. But the project I actually think about in the shower is a machine that turns any novel or biography into a film with one click.
Full Stack
Database to design system
1 click
The target for novel → movie
∞
Novels & films still queued
Frame 01 · Davy Chen
China
About
A developer with a
A developer with a
cinema in his head
- NameDavy Chen
- RoleFullStack Developer
- Based inChina
- FocusWeb apps, pipelines, AI tooling
- Off-screenNovels & movies
- StatusBuilding the impossible thing
I'm Davy. I ship full-stack products — the kind where you own the whole path from a database schema to the last pixel of a hover state, and the CI job that puts it all in front of real users.
Most of my day is ordinary and satisfying: modelling data, designing APIs that don't surprise anyone, making interfaces that feel effortless, and quietly deleting more code than I add. I like systems that are boring in production and delightful to use.
The rest of my life happens in two places. With a novel in hand — the longer, the denser, the better. Or in front of a film, rewinding a scene to work out how it was built.
Those two halves eventually refused to stay separate, which is how the dream below started taking up all my side projects.
Craft
What I do all day
Four layers, one developer. I'd rather understand the whole stack than be precious about any single layer of it.
Backend
APIs and data models that stay honest under load — plus the async plumbing that keeps long jobs moving.
Infra & Delivery
Containers, pipelines, and observability. If deploying is scary, the problem is the process, not the code.
AI & Media
Where the day job starts bending toward the dream: LLM pipelines, prompt orchestration, and media processing.
How I work
Small commits, obvious names, tests where they pay for themselves, and documentation written for the person who inherits the repo. I like prototypes that get thrown away and systems that outlive me. When something is genuinely hard, I'd rather say so early than pretend it's a two-day job.
Passions
The two things I never got tired of
Everything below the surface of this page comes from these two habits.
Novels
Novels are the format with the most room in it. A film gives you two hours; a novel gives you a decade, a family, a country, and the private thoughts of everyone inside it. I read for the parts that can't be filmed — and then I start wondering how I'd try anyway.
Currently reading: whatever keeps me up past midnight. Currently watching: the adaptation, which I will inevitably compare to the book.
Movies
Film is where I learned about structure without knowing that's what I was learning. Pacing, reveals, the exact second a score enters — all of it is engineering you can feel. Half of what I know about good software I stole from good editing.
Award-season completionist, reluctant subtitle hater, and the person who stays through the credits for the score.
The Dream
One click, novel → movie
I believe every classic novel and every great biography already contains a film. The script is in there, the shots are in there, the performances are in there — waiting to be extracted. My dream is a pipeline that does the extraction, so anyone can watch the books they love instead of only reading them.
01Ingest
02Understand
03Adapt
04Design
05Generate
06Assemble
Drop in a book
Plain text, EPUB, or a scanned edition. Any language, any length — a 900-page biography is just a longer input, not a different problem.
Build a story graph
Characters, relationships, locations, timelines, and the emotional curve of the whole book — extracted once and reused by every later stage.
Cut it into a screenplay
Beats become scenes, scenes become industry-format script pages with dialogue, action, and a runtime budget. This is where the hard editorial judgement lives.
Direct it before shooting
Shot lists, storyboards, character sheets, wardrobe and location bibles, casting voices, and a score direction — a production plan, not a prompt.
Render the shots
Image and video models generate each shot from the boards, keeping faces, wardrobe, and light consistent across scenes. Text-to-speech performs the dialogue.
Deliver the film
Automated edit, sound mix, subtitle timing, and colour pass. Out comes a finished cut with a title card — not a slideshow, not a summary. A film.
I believe my dream can come true very soon.
Davy Chen · on the novel-to-movie pipeline
Roadmap
Where the dream stands
A firm belief is not a plan. So here is the plan — roughly in the order the problems actually have to fall.
Now
In progress- Reading and parsing full-length novels without losing the thread
- Chapter-level segmentation and summarisation that respects structure
- Character and relationship extraction into a queryable story graph
Next
Hard part- Beat-to-scene conversion with real editorial rules and runtime targets
- Character consistency across every shot — the make-or-break problem
- Voice casting and dialogue performance that doesn't sound generated
Later
The goal- Automated editing, scoring, and a full sound mix
- Region-aware, culturally accurate adaptation choices
- One click: title in, film out.
Contact
Want to talk about
Want to talk about
books, films, or code?
I'm always up for a conversation with people who think books and movies should be closer together — or who want help building the pipeline. Reach out.
