Ferhat Rudvanoğulları

Data engineer at SupplyBridge, İstanbul

I build the pipelines that turn half a million supplier websites into clean, searchable data for an automotive sourcing platform: crawling and classifying pages, normalizing addresses, matching duplicates, and running LLMs on our own GPU where rules aren't enough.

I studied mechatronics engineering at Yıldız Technical University. That work lives on the engineering page.

Experience

Data Engineer, SupplyBridge

Mar 2026–now

I own the data side of a live automotive supplier database on AWS and GCP: Common Crawl ingestion, address normalization, entity resolution, LLM enrichment on a self-hosted GPU, and safe bulk imports into production PostgreSQL. I also ship product features, CI/CD and monitoring for the app team.

  • Python
  • TypeScript
  • PostgreSQL
  • AWS
  • GCP
  • SGLang
  • Docker

Selected projects

Located half a million suppliers' websites in the Common Crawl archive and classified their pages into seven categories: 485,562 domains in under 25 minutes.

1.08M pages in 7 categories

Turned raw supplier pages into structured profiles with an open-source LLM running on our own GPU, and more than doubled its throughput.

2 → 4.8 domains/s, 1 process

Replaced a paid AI call per address with a self-hosted ML address parser (libpostal) and took throughput from 100 to over 15,000 addresses a minute.

$0 model cost per address

Turned a rule-driven procurement agent into a fixed pipeline behind a private API on Cloud Run, with a predictable output, capped model calls and evidence for every claim.

≤ 2 model calls/company

All 16 projects

Engineering

The levitation test rig: motor controller, battery and wiring mounted on an aluminium frame, with an emergency stop button

Before data, mechatronics. Three seasons on the Starloop hyperloop team: Türkiye's first hyperloop levitation in 2022, second place at TEKNOFEST 2024. Then a levitation rig I built and controlled with a neural network for TÜBİTAK 2209-A. See engineering work