Undrafted Analytics
NFL Analytics Podcast
German football analysis powered by data science
1,000+
Simulations
Parallel game predictions per matchup
40+
State Nodes
Graph-based NFL game flow model
22+
Episodes
Published during 2021 NFL season
5
Seasons
Historical data (2016-2021, 32 teams)
The Problem
German NFL fans lacked data-driven analysis in their native language. Traditional sports predictions rely on simplified models that don't capture the full complexity of football games - play types, situational factors, and team-specific tendencies. There was no platform combining rigorous statistical analysis with accessible German commentary.
The Solution
Created a two-part technical platform: (1) A Monte Carlo simulation engine that predicts NFL game outcomes by simulating complete games play-by-play using a graph-based state machine with 40+ nodes modeling all play types, and (2) A JAMstack podcast website with data visualizations, full-text search, and podcast integration. The simulation uses KDE-based probabilistic sampling with defense-adjusted factors from 5 seasons of historical data.
Tech Stack
Simulation
Data
Frontend
Infrastructure
My Role: Co-founder, Co-host & Technical Lead
- Co-founded German NFL analytics podcast
- Designed and built Monte Carlo simulation engine from scratch
- Created graph-based state machine with 40+ nodes modeling NFL game flow
- Implemented KDE models for probabilistic play outcome sampling
- Built ETL pipeline for PostgreSQL historical data (5 seasons, 32 teams)
- Developed Vue.js/Gridsome podcast website with dark mode and search
- Co-hosted weekly podcast episodes with data-driven analysis
Key Differentiators
End-to-End Data Science Platform: From raw NFL data to published predictions
DOT-Based Game Definition: Entire NFL game flow in human-readable Graphviz format
Novel Defense Factor: KDE-based team vs league defensive distribution comparison
German Market Niche: First German-language NFL analytics podcast with rigorous data science
Want to discuss this experience?
I am open to full-time Senior AI/ML Platform Engineer roles where this kind of production AI, data, and platform work is useful.
Other Projects
Enterprise IoT Platform
Multi-Tenant Workflow Automation Infrastructure
Reduced customer onboarding time by 70% and eliminated 40+ hours/month of manual DevOps while the architecture was designed for 1000+ isolated workflow instances.
Zeitgaist
Cross-Lingual Social Intelligence
Synthesizes current insights from 6 platforms in seconds vs hours of manual monitoring, with cross-lingual search and filterable source context.