SkillPay AI--From skills to earning

Students have real skills, but turning them into income is hard. Three problems kept coming up:

Nobody tells beginners what to charge. They either sell 30 hours of work for $5 or quote a number that scares clients away.
Vague briefs turn into unpaid work. A request like "make me an app like Instagram" has no limits, so the extras never stop.
Students fear not being paid. After delivering source code, a client can simply disappear.

We wanted one platform that fixes all three, so a student's first paid project feels safe instead of risky.

What it does

SkillPay AI takes a student from "I have a skill" to "I got paid":

Smart Rate Engine. It looks at skill category, experience tier, complexity, urgency and revision rounds, then produces a fair quote with a client-ready justification. An anti-exploitation floor of $15–$24/hr stops students from undercharging.
AI Scope Clarifier. It turns a vague request into a formal Scope of Work with 3 milestones (25% / 50% / 25%), flags unbudgeted add-ons as paid out-of-scope work, and exports a contract.
Beginner Marketplace. Projects are pre-funded, so the money is already in escrow when a student applies.
Milestone Escrow. The student submits a GitHub and demo link, and the client approves to release funds. If the client goes silent, a 72-hour auto-release pays the student automatically. Withdrawals go to PayPal, UPI or bank transfer.
Live Database Inspector. Judges can watch the SQL tables change in real time as money moves.

The quote is built from the project inputs and checked against the floor:

$$\text{Quote} = H \times \max\left(r_{floor},\ r_{base} \cdot m_{tier} \cdot m_{complexity} \cdot m_{urgency}\right) + C_{revisions}$$

The total is then split across milestones:

$$P = 0.25P + 0.50P + 0.25P$$

Frontend: a single-page app in HTML and Tailwind CSS with a multi-role dashboard (Student and Client views, account switcher, database inspector).
Backend: a REST API in both Node.js (Express) and Python, so it runs with whichever runtime is installed. Both expose the same endpoints.
Database: a relational SQLite schema with users, projects, project_skills, escrow_contracts, milestones, proposals, transactions and disputes.
Money movement: milestone approval runs as a real SQL transaction. Funds move from escrow to the student's wallet and are recorded in a transaction ledger for auditing.
One-click launchers (start.bat, start.ps1) and an offline mode that falls back to local data, so the demo works even without a server.

Keeping money correct. Releasing escrow has to be all-or-nothing, so we put the balance update and the ledger entry in one SQL transaction.
Two backends, one behavior. Keeping the Node.js and Python servers identical, so the same frontend works with either one, took careful testing.
Pricing that is fair to both sides. A floor protects students, but the quote still has to be defensible to a client, so we generate a justification script with each rate.
Making trust visible. Escrow is invisible by nature, so we built the live ledger and database inspector to let people see where the money is.

Designing financial flows means thinking about failure first: disputes, ghosting clients and double releases.
A clear scope document prevents more conflict than any dispute system.
Showing the data (ledger, tables, statuses) builds more trust than a polished UI alone.

Connect withdrawals to live PayPal payouts (sandbox first, then production).
Smarter scope generation using an LLM.
Student reputation scores based on completed, fully paid milestones.
Automated dispute mediation with evidence from GitHub commits.