All work

05 / Collaborative knowledge platform

WikiMasters

A collaborative wiki MVP for creating, editing, and reading articles. Authored knowledge stays central, supported by media, cached reads, and AI summaries.

Role
Full-stack engineer
Period
2026
Focus
Content · caching · media · AI
Status
MVP

01 / Product challenge

An article connects identity, storage, and AI.

Creating and reading articles is the visible experience. Behind it, identity, stored content, cached results, media, and AI summaries each have a different responsibility.

An authenticated author creates an article with tags and optional media. Readers browse the list, search titles and summaries, and open the full article. Editing regenerates the summary and invalidates the cached list.

One article / clear responsibilitiesContext coordinated
Identity

Who is using the product

Persistence

Where authored knowledge lives

Delivery

How readers retrieve content

Enrichment

What helps readers understand it

02 / Content lifecycle

One source of truth. Several ways to deliver it.

The article remains the reference point. Cached reads, media, and summaries support different parts of the reading experience.

Authored content

The article

Created, edited, and stored as a durable record.

Read efficiently

Reuse cached results while keeping the database authoritative.

Keep media connected

Reference uploaded assets alongside the article.

Offer an overview

Provide a generated summary as an additional reading aid.

03 / Engineering decisions

Three boundaries keep the knowledge dependable.

Each integration answers a product need and introduces a responsibility that the application must own.

01 / Stored record / cached copy

Fast reads, with an authoritative record.

Pressure
Readers can request the same content repeatedly.
Decision
Cache the article list in Redis for 60 seconds. Creation and edits delete that list key; article detail reads query Postgres directly.
Cost
The TTL bounds list freshness; invalidation adds a dependency on Redis.
02 / Article record / media asset

Store the article and its media deliberately.

Pressure
Article records and uploaded files have different storage needs.
Decision
Keep relational metadata in the database and media in object storage.
Cost
Two storage systems need coordinated references and lifecycle handling.
03 / Authored content / derived summary

Keep summaries downstream of the source.

Pressure
A short overview helps readers orient themselves, but generated text is supplemental.
Decision
Generate a summary from the title and content during creation and editing, then store it alongside the article for list and search previews.
Cost
Generation adds a service dependency to writes; readers still need the authored article as their reference.

04 / Product scope

A working wiki, with room to grow.

The project is an MVP. Its current foundation covers the article experience and supporting services; deeper collaboration remains a future direction.

Current foundation

Author, organize, and read.

  • Article creation, editing, and listing
  • Authentication with Stack Auth
  • Typed database access and Redis caching
  • Media storage and AI article summaries
Future directions / repository ideas

Deeper collaboration.

  • Revision history and change comparisons
  • Live editing with presence
  • Comments and discussions
  • Draft, review, and publishing workflows

Scope follows the project README; future ideas are not presented as shipped features.

05 / Outcome

Create and edit an article, attach media, and help readers browse it through tags and AI summaries.

Authenticated authors create and edit articles with tags and media. Readers browse and search titles and summaries, then open the full article. The list cache expires after 60 seconds and is invalidated after writes.