Project Overview
In the fast-paced world of sports media, creating engaging fight analysis content is both time-intensive and expensive. Professional video production—from data collection to scriptwriting to final video rendering—can take content creators 6-8 hours per fight and cost upwards of $500 when factoring in analyst time, scriptwriters, and video editors. For sports media companies covering multiple fights per event, this becomes operationally unsustainable.
I engineered an end-to-end intelligent automation system that transforms raw UFC fight data into publication-ready AI video analysis in minutes—reducing production time by 90% while maintaining broadcast-quality output.
This project showcases my expertise in web scraping, data engineering, AI workflow automation, and API orchestration—solving a real-world content production bottleneck that affects sports media outlets, betting analysis platforms, and UFC content creators globally.
Sports media companies and UFC analysts face three critical challenges:
1. Data Fragmentation
Fight statistics are scattered across multiple platforms (Sherdog for play-by-play commentary, UFCStats for quantitative metrics). Manual data collection from these sources is tedious, error-prone, and time-consuming.
2. Content Creation Bottleneck
Professional fight analysis requires domain expertise to synthesize technical data into compelling narratives. Hiring scriptwriters and video editors for every fight creates unsustainable costs and turnaround times—especially during events with 10+ fights.
3. Scalability Limitations
As content demand grows (pre-fight analysis, post-fight breakdowns, highlight reels), traditional production workflows break down. Teams burn out, miss publishing windows, and lose competitive edge to faster publishers.
The result? Delayed content, missed revenue opportunities, and an inability to scale coverage across multiple fight cards.
I built a webhook-triggered automation system that orchestrates web scraping, data transformation, AI script generation, and video synthesis—all running autonomously on Make.com.
1. Centralized Control Hub
Google Sheets serves as the command center where users input fight URLs and trigger automation by updating a "Status" field. This design allows non-technical team members to generate videos without touching code.
2. Intelligent Web Scraping Engine
Dual-source data extraction: Built custom HTTP scrapers for Sherdog (play-by-play narrative) and UFCStats (quantitative fight metrics)
Dynamic HTML parsing: Engineered robust parsers to handle JavaScript-rendered content and inconsistent HTML structures across both platforms
Data normalization pipeline: Transformed disparate data formats into standardized JSON schemas for downstream processing
3. AI-Powered Script Generation
GPT-4 Integration: Leveraged OpenAI API with custom prompt engineering to generate contextually rich fight analysis
Adaptive narrative styles: Configured the system to produce multiple content formats—hype videos, technical breakdowns, or highlight summaries—based on audience and platform requirements
Data-driven storytelling: The AI synthesizes numerical stats (strike accuracy, takedown percentages) with play-by-play action to create compelling narratives that resonate with fight fans
4. Automated Video Synthesis
HeyGen API integration: Connected Make.com to HeyGen's avatar video platform via API
Customized AI presenters: Configured virtual avatars with professional voiceovers to deliver scripts as broadcast-quality video content
One-click rendering: From data scraping to final video export—fully autonomous
5. Quality Assurance & Logging
Real-time status tracking: Automated logs update in Google Sheets after each pipeline stage, providing transparency and error traceability
Webhook-driven execution: Zero manual intervention required—system activates instantly when status changes
Solved complex data normalization challenges by converting unstructured HTML tables and narrative text into structured JSON objects optimized for AI consumption. This required regex pattern matching, conditional parsing logic, and schema validation to ensure data integrity.
Seamlessly integrated multiple third-party APIs (OpenAI, HeyGen) within Make.com's visual workflow builder, managing authentication, rate limiting, and error handling across distributed services.
Implemented conditional branching in Make.com to route different fight types (title fights vs. prelims) through specialized script templates, ensuring contextually appropriate narratives.
⚡ 90% reduction in production time
What previously took 6-8 hours now completes in minutes—enabling same-day publishing for time-sensitive fight analysis.
💰 Estimated cost savings: $450+ per video
Eliminates the need for dedicated scriptwriters ($150) and video editors ($300+) while maintaining professional output quality.
📈 Infinite scalability
The system handles 1 fight or 50 fights with identical efficiency—no additional headcount required.
🎯 Reusable infrastructure
The automation blueprint extends beyond UFC—adaptable to boxing, MMA promotions, or any combat sports with structured data sources.
Here is the github link, where the python code is available for public if anybody need the coding reference.