> initializing portfolio_

BenediktVennen

>_

Turning raw data into actionable insights. Building intelligent systems that automate, predict, and explain.

Frankfurt am Main, Germany
bvn3141@portfolio ~ zsh
SCROLL

01 / about

About Me

Benedikt Vennen

Job titles, frankly, have always felt a bit constraining. Like wearing a suit two sizes too small. Functional, sure, but not exactly how you'd choose to move through the world. I'm a problem solver. That's the actual job description. Whether the solution involves a Python script, thorough research, a conversation no one else wanted to have, or a completely unconventional approach nobody thought to try. I'll find the lever that moves the thing.

Two things genuinely light me up. First: patterns. Not the obvious kind. Those are boring. I mean the ones hiding three layers deep in a messy dataset, the correlation that only reveals itself after you've asked the right question at the right angle. Finding something that wasn't supposed to be there, or disproving something everyone assumed was settled fact. That's where I come alive. Second: automation. There is a very specific kind of satisfaction that comes from watching a process that used to cost someone three hours of their week run silently in the background while they go get coffee. I build that. Repeatedly, and with genuine enthusiasm.

I do my best work in the spaces between departments, those cross-functional conversations where nobody's quite sure whose problem it actually is. Spoiler: it's usually shared, and shared problems are the interesting kind. I like helping people make decisions they couldn't fully justify before. Not because the data told them what to do, but because it finally showed them what was actually going on. I identify weaknesses in causal chains. I poke holes in logic that has already been signed off on. I illuminate things that were sitting comfortably in the dark. Not out of stubbornness. Purely out of curiosity.

As for tools: I don't have a favourite hammer. The problem gets to decide what gets thrown at it. If the most promising approach is something I haven't used yet, I'll be dangerous with it before the week is out. Modern technology moves fast. I find that exciting rather than exhausting. New framework, new LLM workflow, entirely new domain: I'll find my footing, and I genuinely enjoy the process of getting there.

3+
Years Experience
20+
Projects Delivered
3
Domains of Expertise
∞
Data Curiosity
QUICK_INFO
name:Benedikt Vennen
role:Data Analyst / Data Scientist / Sports Scientist
location:Frankfurt am Main, DE
status:Open to opportunities
email:b.vennen@gmx.de

02 / skills

Technical Skills

🐍

Python Ecosystem

The daily driver. From scraping PDFs and web data to wrangling messy datasets, building statistical models and producing visuals that don't make people question your taste.

pandasnumpymatplotlibseabornscikit-learnstatsmodelsbeautifulsouppdfplumber
πŸ—„οΈ

SQL

Everything pandas does, but at database level. ETL pipelines, CTEs, multi-source joins, aggregations and enough window functions to make a DBA mildly impressed.

CTEsETLWindow FunctionsMulti-source JoinsAggregationsSubqueriesPostgreSQLSQLite
🧠

Machine Learning

Supervised and unsupervised. Classification tasks, dimensionality reduction, and making sense of clusters that weren't labelled to begin with.

Decision TreesRandom ForestK-Nearest NeighborK-Means ClusteringPCAClassificationFeature EngineeringModel Evaluation
πŸ“

Statistics

The part where the math earns its keep. Designing experiments properly before running them, then making sure the results actually mean what they seem to mean.

Hypothesis TestingT-TestsF-TestsPower AnalysisSampling StrategiesProbabilityConfidence IntervalsDescriptive Stats
βš™οΈ

Automation

If it's repetitive and rule-based, it probably shouldn't involve a human. PDF parsing pipelines, scheduled scrapers, organisational tooling built around real workflows.

PDF PipelinesWeb ScrapingpdfplumberScheduled ScriptsETL AutomationOrganisational AppsAPI IntegrationProcess Design
πŸ€–

Agentic AI

Context engineering more than prompt engineering. Building agent workflows that handle research, drafting and processing tasks autonomously β€” so the boring work gets done while I focus on the interesting parts.

Claude APIAgent WorkflowsPrompt EngineeringRAG SystemsContext DesignLLM PipelinesResearch AutomationEmail Drafting Agents

04 / publications

Publications

PUBLICATIONS_LOG

> loading publications...

> status: pending

> section coming soon_

Scientific publications and research papers will be listed here.
Data will be populated soonβ„’