Combating Internet Brain Rot Through AI
Advanced AI-powered content moderation research to foster healthier digital environments. Open-source models, comprehensive datasets, and rigorous academic methodology.
Advancing Digital Wellness Through AI Research
Founded by computer science students at FEU Institute of Technology, we're developing cutting-edge solutions to combat the degradation of online discourse and critical thinking.
Research-Driven Approach
Our interdisciplinary research combines deep learning and behavioral psychology to understand and address the root causes of neurodegenerative online content.
Through rigorous academic methodology and open-source development, we're building transparent, accountable AI systems that can be audited and improved by the global research community.
Transfer Learning
Augmented tools built on top of industry-leading models
Content Protection
Moderation of harmful, explicit, and uneducational content
High-Performance Computing
Trained and hosted on powerful Google Cloud TPUs & NVIDIA GPUs
Precision Filtering
High accuracy, customizable, and critical content analysis
Upload Your Video
Demo with our fine-tuned multimodal model specializing in recognizing sludge videos - a subtype of internet brain rot that grabs attention by presenting mutliple streams of visual clips (e.g. Subway Surfers, Family Guy, Soap Cutting,)
Drag and drop your video here
or click to browse files
Video preview will appear here
Supported format: MP4 only (Max size: 50MB, Max duration: 1 minute)
Advancing AI Content Moderation
Our research spans multiple domains with a focus on practical applications, open-source development, and academic rigor.
Visual-Qwen: Augmenting Multimodal Deep Learning with Attention Mechanisms to Recognize “Sludge” Videos from Short-Form Content
FEU Institute of Technology - Computer Science Thesis
- • Open-source multimodal dataset of 2,000 short-form videos (6k paired video/audio/text rows), human-model-annotated and expert-validated for sludge vs. non-sludge classification
- • Novel multimodal architecture (CLIP ViT-G/14 + Whisper V3 Turbo + Q-Former + Qwen3-4B) leveraging cross-modal attention for video content moderation
- • Two-stage training pipeline (linear-projection pre-training on LLaVA image-caption pairs + LoRA fine-tuning on custom sludge dataset)
- • 93.50% accuracy, 91.09% precision, 95.83% recall, 93.40% F1 on held-out test set
- • High acceptance validated under TAM and ISO/IEC TR 24028 standards with content creators, moderators, and ML experts
- • TPU v4-64 training methodology at FEU Institute of Technology via the TPU Research Cloud
Development Timeline
From research inception to commercial launch - tracking our milestones and future roadmap.
Startup Founded
MicroMarc Research officially established
Research Inception
Initial research and development of the MicroMarc AI system
Initial Dataset & Grant
Ethically scraped 2K dataset + TPU Research Cloud grant secured
Novel Architecture
Developed breakthrough multimodal AI architecture
Model Training
Trained initial multimodal content moderation model
Expert Validation
Verified with 10 AI/Data experts & 40 content creators/moderators
Meet the Researchers
Four dedicated CS students and their extraordinary advisor from FEU Institute of Technology, combining academic excellence with entrepreneurial vision.
Justine Jude Pura
Project Mentor
Marc Olata
Founder & CEO
Alpha Romer Coma
Founding Engineer
Job Isaac Ong
Founding Data Scientist
Kristoffer Ian Sioson
Founding Data Scientist
Target Audience & Business Model
Addressing multiple market segments with AI-powered content moderation solutions.
Social Media Platforms
Enterprise-grade content moderation APIs
Parents & Families
Mobile app for child digital wellness
Content Creators
Specialized content analysis & recommendations
Content Moderators
AI-powered content moderation tools
Digitally Conscious Users
Background agents for healthier browsing
Revenue Streams
Business Model
Sponsors & Partners
Grateful for the support of organizations sharing our vision of healthier digital environments.
YouTube Researcher Program
Ethical Data Collection
TPU Research Cloud
AI Infrastructure
🤝 Become a Partner
Support cutting-edge AI research and help combat internet brain rot through strategic partnerships.