CASE STUDY 16 / Personal project · Full-stack AI storytelling

Love Story Theater

I made this for a friend going through a breakup: a conversational app that turned shared memories and relationship details into hopeful fictional stories. I built a Next.js frontend and a separate Kotlin / Ktor backend connected to DeepSeek. The service is now offline and the project is no longer maintained.

Offline · No longer maintained

CONTEXT

Personal project

MY ROLE

Independent full-stack developer

TOOLS & FOCUS

Next.js / React · TypeScript / MUI · Kotlin / Ktor · DeepSeek API · i18n

01 / THE PROBLEM

What needed to work.

The idea began with a friend’s breakup. I wanted to make something personal and comforting: a conversation that gathered details gradually, then used them to write an imaginative, hopeful story. The engineering task was to connect that guided chat experience to an AI service through a separate backend.

CONCEPT / 16

01Guided Next.js conversation

02Kotlin / Ktor JSON API

03DeepSeek-generated story

Illustrative workflow · Not a product screenshot

A summary of my responsibilities and technical approach. This workflow illustration does not reproduce the product UI.

02 / MY CONTRIBUTION

Where I took ownership.

Built the Next.js, React and TypeScript chat interface with Material UI, message history, typing feedback and a story-generation loading state.
Implemented a separate Kotlin / Ktor backend with JSON endpoints for contextual follow-up questions and final story generation.
Connected the backend to DeepSeek and wrote prompts for the conversational tone, supplied context and requested story language.
Organized English, Chinese and German interface copy and question pools separately from the chat rendering.

03 / TECHNICAL APPROACH

Decisions in context.

Separate the interface from model calls

The Next.js client handled the conversation and called a configurable backend URL through Axios. The Kotlin / Ktor service accepted JSON requests, built prompts and called DeepSeek. This kept the browser interface and model integration in separate applications.

04 / DELIVERY & REFLECTION

What I can share.

Delivery.

A personal full-stack experiment, originally made for a friend after a breakup. The service has stopped running and the project is no longer maintained. The source repository remains available to show the separate frontend and backend implementation.

Have a similar challenge?

Let’s talk ↗