Independent project · 06/2019 — 12/2022
Steam BOT
An automated marketplace intelligence and trading operation that monitored more than 300,000 game items per day, identified pricing opportunities and supported purchases and sales across Steam and BitSkins. It completed 16,278 sales and generated approximately R$90,000 in estimated net profit over more than three years.
01
Problem
Trading decisions required frequent collection and analysis of a marketplace catalog too large for manual monitoring.
02
My responsibility
I built and operated the web scraping, ETL and analysis pipeline, including resilience for long-running collection.
03
Architecture
Python and Selenium workers collected marketplace data through managed proxies, feeding an ETL flow used by automated trading rules.
04
Technical challenges
High-volume collection, source instability, proxy management, data consistency and reliable daily execution.
05
Results
The system extracted more than 300,000 items daily and the account recorded 16,278 completed sales. Based on my operating records, it generated more than R$2,500 in mostly automated net monthly income and approximately R$90,000 in estimated lifetime net profit. The estimate accounts for Steam and game fees of roughly 10–15% and cash conversion near 87% through BitSkins.
Operational evidence
