HomeWorkCase study 01 · 2025

Sneaker deals · Automation

Deal tracking that handles 2,500 operations a day

Watch several marketplaces at once, score every find, push the hits straight to a phone — without anyone keeping tabs open.

The list, minute by minute16:9

2,500

operations per day · one VPS

Year
2025
Client
Own product
Role
Sole developer · concept to operations
Areas
Full-stack development · Cloud & Deployment
Screenshot · One hit as it reaches a phone
Where it starts: the alert21:9
Screenshot · VPS load across a single day
Where it ends: the load on one VPS21:9

The starting point

Resellers have to watch several marketplaces at once. Notice too late that an offer is online and it's gone. Doing that by hand isn't possible.

What was built

The list, while it fills16:9

A full-stack TypeScript platform that continuously scrapes several marketplaces, scores every find automatically with AI, and pushes relevant hits out in real time via Telegram and web push. Updates reach the frontend over WebSocket, without reloading the page.

It all runs on a single VPS: scraper, scoring, WebSocket, push.

  • 01What you touch

    Screenshot: the overview with filters and scoring
  • 02Where the data sits

    Screenshot: an alert as it arrives
  • 03How it is operated

Stack

Interface

Next.js React WebSocket

API

TypeScript Cheerio DeepSeek

Data

PostgreSQL / Prisma

Cloud

PM2 VPS

Result

Live since 2025: 2,500 operations a day, CPU at 34 percent.

New items land at the top and the rest shift down. No reload.16:9
View · The deal list
The dense table a user actually looks at.16:9

snipeflip.com

The other work

The three other case studies.

Screencast · RecRobot: from 148 approached to 9 hired

03Recruiting · Automation

3 → 12

developers in three months

RecroBot

All projects

Contact

Half a minute, and you know more.

Half a minute on what I work on and how. If a question is left, write to me — an answer within one working day.