Memory Lantern 🏮: Stories That Stay — Preserving What Makes Us Human with AI
My Week 4 project for the MLH x DEV Writing Challenge
Four weeks. Four projects. Four different ways to explore how technology can make everyday life better.
For the final week, I wanted to build something different. Not another productivity tool, business assistant, or health dashboard.
I wanted to explore something deeply human: our memories and the stories we leave behind.
What I Built
I built Memory Lantern — Stories That Stay, a memory garden within the Dr. T platform.
Memory Lantern helps people preserve family stories, recipes, traditions, meaningful sayings, photographs, and moments of everyday joy.
The idea was inspired by a question:
What if AI could help us remember without pretending to be the people we miss?
For Hacktoberfest Week 4's "Ghost in the Machine" theme, I chose a gentler interpretation of the ghost: not a digital imitation of someone who has passed away, but the traces people leave behind through their stories, voices, customs, and acts of love.
Each memory becomes a lantern — something that can be revisited, organized, and shared intentionally.
The experience includes:
Memory Garden: Preserve meaningful memories in one place.
Timeline: Explore memories chronologically.
Storybook: Bring selected memories together into readable stories.
AI-assisted narratives: Transform selected memories into editable stories and tributes.
Printable remembrance pages: Create keepsakes that can be shared.
Archive and export: Keep personal stories accessible beyond the application.
Privacy and human control are central to the concept. Fictional sample memories allow people to explore the experience without exposing private family histories.
My guiding principle is:
Preserve the memory, not impersonate the person.
Demo
Video: https://youtu.be/qYLhiOIKh5Q?si=2Cb7wayD0nNabdIn
Live preview:
https://ais-dev-4s4jvpipr3mh3mz6x2hpfp-393352619239.asia-southeast1.run.app
Open the Memory Lantern tab in Dr. T.
GitHub repository:
https://github.com/Zenieverse/Dr.-T
DEV.to project story:
https://dev.to/zenieverse/memory-lantern-stories-that-stay-n3k
The interface and integration are implemented. One important limitation: the hosted preview reported that the Gemma inference provider was not configured, so successful live Gemma story generation is not yet verified. The AI-generated workflow should therefore be considered an integration target until a real generation succeeds.
Partner Technologies
The main open-weight AI technology explored in this project is Google Gemma 2.
I designed the integration around a server-side provider adapter supporting local Ollama inference or a compatible hosted inference endpoint.
The intended workflow is:
A user selects memories they want to preserve.
The application sends the selected material to a server-side story-generation endpoint.
Gemma generates an editable narrative when the provider is configured.
The user reviews, edits, and decides whether to save the story.
I also implemented safeguards against inventing biographical facts or impersonating deceased people.
Other technologies include:
TypeScript: Application logic and domain models.
React: User interface and interactive memory experiences.
GitHub: Source control and implementation tracking.
Ollama-compatible inference architecture: An option for running open-weight models locally.
What interested me most about open-weight AI was the flexibility it offers. Developers can experiment with model hosting, inspect their implementation, and explore different privacy trade-offs.
For personal memories, that flexibility matters.
Hackathon Experience
The MLH x DEV Writing Challenge has been an exciting four-week building journey.
Each week pushed me toward a different problem and a different perspective on technology.
Week 1 — Pianist Touch Grass
I explored how real-world sounds could inspire personalized piano learning. It combined creativity, music, nature, and AI.
Week 2 — MicroBiz
I moved into practical AI for entrepreneurs and small businesses. MicroBiz brought together learning, marketing, business planning, and receipt/document processing using an open-weight AI approach with Gemma 4. It has an upgrade now post the submission as the next version with Payflow DebtFree module on the same MicroBiz tab on the same Dr. T platform.
Week 3 — HealthAlert Companion
I explored health technology and responsible AI-assisted health information. The goal was to help people understand alerts and identify appropriate next steps without replacing professional medical care.
Week 4 — Memory Lantern
I returned to something personal: preserving memories, family traditions, and stories that matter.
Looking back, these projects represent four different dimensions of human life: creativity, livelihood, health, and connection.
The biggest lesson was that AI is most meaningful when it supports people rather than replaces their judgment, creativity, or relationships.
I also learned that building responsibly means being honest about technical limitations, testing real integrations, and making privacy part of the design rather than an afterthought.
What I Learned
This challenge reinforced three lessons:
A good AI project starts with a human problem, not a model.
Open-source and open-weight technologies create opportunities for experimentation, transparency, and community improvement.
A working interface is not the same as a verified production integration. Testing and honest documentation matter.
What's Next?
I want to continue improving Memory Lantern, particularly its model integration, privacy controls, multilingual storytelling, and export experience. In that for the community spirit, I have launched a StreetvendorsMap inspired by ganhmuusinh.vn, my clone is an English/Vietnamese plus some upgrades hosted on the same Dr. T platform; and ...
I hope it can eventually help families preserve the little details that might otherwise disappear: a grandmother's recipe, a parent's favorite saying, a childhood story, or a family tradition passed between generations.
A machine cannot replace the people behind those memories.
But perhaps it can help us preserve their stories.
A lantern does not replace the light that came before it. It helps us find our way back to it. 🏮
Thank you to MLH and DEV for four weeks of learning, building, experimenting, and sharing.
Top comments (0)