DragSet’s social mechanics were only useful if they stayed specific to car enthusiasts. I worked with vehicle profiles, race availability, a swipe deck, and the implemented race-challenge action instead of flattening the experience into generic profiles and matches. That product-language decision makes the workflow easier to understand and the portfolio evidence more credible. Takeaway: model the domain’s real objects before choosing familiar UI patterns.

Title: AI-Native Full-Stack Engineer for Private Portfolio Intelligence App

I’m building ThesisFrame, a private, single-user portfolio intelligence app. It is not a trading or signal product: it helps the user review what changed in a portfolio, connect evidence to an investment thesis, and clearly surface missing or stale data.

The existing project already includes:

TypeScript / Node.js intelligence service

PostgreSQL

React-based Wealthfolio add-on

Docker / Compose deployment setup

Written architecture and phased development plan

I’m looking for one strong, pragmatic full-stack engineer who uses AI tools productively but still ships tested, understandable, maintainable code.

Initial paid trial — 8 to 12 hours: Build a small read-only adapter from Wealthfolio portfolio snapshots into the Intelligence service.

Deliverables:

Clean TypeScript implementation

Fixture-based test data

At least one automated test

Clear README / run instructions

One focused GitHub pull request

No access to real portfolio data, credentials, or production infrastructure

If the trial goes well, this can become a part-time engagement of roughly 10–20 hours per week, delivered through small milestones.

Required experience:

Strong TypeScript, Node.js and React

PostgreSQL schema design and migrations

REST/API integrations, rate limits, retries and error handling

Docker / Docker Compose

Automated testing and GitHub PR workflow

Comfortable reading an existing codebase and following a written architecture

Strong plus:

Financial-data APIs, data provenance, ETL/ingestion pipelines

Hono, Zod, React 19, scheduled jobs

Experience building with Cursor, Codex or Claude Code while retaining engineering discipline

Important constraints:

Every external data point needs source, timestamp and freshness information.

Missing or failed data must become an explicit “Data Gap”, never a silent zero or “no change”.

No automated trading, Buy/Sell/Hold advice, price targets or execution features.

All work must be committed to my GitHub repository. I retain full IP ownership.

Please do not apply if your approach is primarily unreviewed AI-generated code.

When applying, please send:

One relevant GitHub repo or PR you personally wrote.

A short explanation of how you would approach the paid trial.

Your hourly rate and weekly availability.

A brief example of a production bug you caught with a test.

Budget: Please propose an hourly rate for a paid trial first. I value quality, communication and test discipline over the lowest price.