NiyonshutiDavid/uruti-platform

Uruti is an AI-driven pre-acceleration platform designed to bridge the early-stage "Valley of Death" for startups in Rwanda and emerging ecosystems. # URUTI AI-powered ideation, pitch coaching, and investor intelligence platform for early-stage founders. ## Project Links - Current development repo: github.com - Previous versions: github.com ## Deployment And Deliverables - Deployed web app link: uruti.rw - APK files folder: drive.google.com - Demo video (5 minutes): drive.google.com ## Hugging Face Model Deployments The platform models are deployed on Hugging Face: - Pitch coach model: huggingface.co - Investor ranker and analyzer model: huggingface.co - Chatbot model (GGUF, CPU-friendly): huggingface.co - Chatbot model (Tensor/large variant): huggingface.co ### Investor Ranker And Analyzer (What It Scores) Model family and deployment bundle: - Inference pipeline: XGBoost-based `Pipeline` (loaded from `uruti_bundle.joblib`) - Classes: - `not_ready` - `mentorship_needed` - `investment_ready` What it analyzes at inference time: - Structured model features (10 canonical fields): - `funding_total_usd` - `funding_rounds` - `founded_year` - `rd_spend` - `administration_spend` - `marketing_spend` - `profit` - `state_code` - `category_code` - `source` - Uruti readiness heuristics used for calibrated score blending: - team strength (`team_size`) - traction (`user_base` vs target market) - growth (`monthly_growth_pct`) - partnerships - sector boost - startup stage - optional platform score (`mlp_score`) when available Score bands used in output: - High Readiness (80-100) - Strong Seed Readiness (70-79) - Mentorship-to-Seed Transition (60-69) - Mentorship Needed (50-59) - Early Stage / …