Hi, I'm Fabian Marin — @fmgblackwolf.

I build full-stack systems and data pipelines for companies ranging from Oracle to early-stage startups. My focus is on clean architecture, ownership, and delivering features end-to-end.

Currently exploring: AI-augmented development workflows (Codex, LLMs) and their practical impact on delivery speed and code quality.

Tech stack: TypeScript, React, Node.js, Java, Python, MongoDB, PostgreSQL, MySQL, Kafka, AWS.

My main project (private): A financial pipeline that processes NASDAQ and NYSE data to calculate fair price valuations.

I work best in autonomous, low-ego environments where shipping matters more than process.

📫 Reach me via LinkedIn

A modular, extensible pipeline that processes all publicly traded companies on NASDAQ and NYSE to calculate fair price valuations and analyze fundamentals.

• Extensible stage-based pipeline: Each pipeline consists of any number of JavaScript/TypeScript stages. Each stage receives company data and enriches it — via API calls, web crawling, financial ratio calculations, or DCF (net-income or FCF) analysis.

• Configurable generator (first stage): Uses JavaScript generators with yield to page through ticker lists in batches. Batch width is configurable to control parallelization. Other pipelines use a different kind of initial stage to query data from a DB such as MongoDB.

• Throttling & rate limiting: Configurable delays between batches to avoid API throttling and respect rate limits from data providers.

• Bulk persistence (final stage): Data is persisted to MongoDB using bulk upserts for efficiency. The design's flexibility easily allows for swapping in a different final stage to store in a different data store, e.g. MySQL.

• Tech stack: TypeScript, Node.js, MongoDB, REST APIs, web crawling utilities.

The full source code is private to protect intellectual property around the valuation models. Select, anonymized code snippets are available upon request to demonstrate specific patterns or implementations.