Six-Month Training Program: “From Steel to Systems”

Recursive AI Foundations: Building Intelligent Systems for Social Impact

Length: 24 weeks (6 months)
Schedule: 3 sessions/week (2 weekdays, 1 Saturday workshop)
Total Hours: ~180 classroom + 80 lab/project hours = 260 hours
Cohort Size: 15–20 participants
Outcome: 1 Capstone Project + 2 Certifications + Portfolio + Public Showcase

Program Breakdown (Month-by-Month)

🟦 Month 1: Foundations of AI & Ethical Computing

Topics Covered:
- History of computing and AI (from Turing to Transformers)
- Introduction to neural networks and basic Python
- AI in Pittsburgh: local and global impact
- Equity, bias, and data ethics in machine learning
Hands-on Projects:
- Build a chatbot with hardcoded logic
- “Bias in a Box” exercise with sample datasets
Guest Speakers:
- Local ethicists, CMU/Pitt professors

🟩 Month 2: Machine Learning Basics + Data Handling

Topics Covered:
- Supervised vs. unsupervised learning
- Dataset structures, cleaning, and labeling
- Introduction to NumPy, Pandas, and Scikit-learn
Hands-on Projects:
- Build a classifier to predict public transportation patterns
- Use Pittsburgh open data for real-world context
Micro-certification:
🏅 “Intro to ML with Scikit-learn”

🟨 Month 3: Neural Networks & RWKV Agent Simulation

Topics Covered:
- How neural nets learn: forward and backpropagation
- RWKV: lightweight RNN-transformer hybrid
- Recursive agent design concepts
Hands-on Projects:
- Train a basic RWKV model with pre-tokenized data
- Launch your first “digital agent” to mimic user input patterns
Special Activity:
- Lab week: simulate agents with different memory parameters

🟥 Month 4: Coding for Intelligence – Python, APIs & Agent Logic

Topics Covered:
- Modular Python coding & Git version control
- Building AI behaviors with APIs (OpenAI, HuggingFace)
- Environment modeling and agent reflex loops
Hands-on Projects:
- Connect a GPT model to a sensor dataset
- Prototype a recursive logic bot that adapts to user style
Group Challenge:
- “Bot vs Bot”: pit evolving agents in conversation to test learning

🟪 Month 5: AI for Good – Social Impact Lab

Topics Covered:
- AI in healthcare, environment, public services
- Building responsibly: Explainability, accountability, transparency
- Community-driven design
Hands-on Projects:
- Group chooses from:
  ▪ Smart trash detection
  ▪ Transit route optimizer
  ▪ Mental health assistant prototype
- Apply prior months’ skills to a real-world local problem

⬛ Month 6: Capstone + Public Showcase

Capstone Structure:
- Final Project: Solo or team-based
- Must include: dataset, model, visualization, ethical analysis
Examples:
- “Fractal Pittsburgh” – agent-based model of city development
- “SteelBot” – explainable AI that interviews displaced workers and suggests retraining paths
Showcase:
Public demo day at the Makerspace (invite funders, mentors, and partners)
Final Certification:
🏅 “Yarian AI Level 1: Responsible System Design”

Additional Components

Element	Description
Mentorship Pods	Each student assigned a mentor (CMU/Pitt students, industry volunteers)
Lab Hours	Evening access to Makerspace tools, cluster compute, 3D printers, and datasets
Ethics Hours	Weekly reflections: bias, data responsibility, explainability
Portfolio Building	GitHub projects, digital resume, final presentation deck

Materials Needed

10 RWKV-capable cluster nodes (or cloud credits)
GPUs with at least 12GB VRAM for model training
Group licenses for: GitHub Pro, OpenAI API, JupyterHub
Curriculum LMS (Moodle or Notion + Discord/Slack for cohort)

Graduation Outcomes

Outcome	Target
Portfolio-ready project	✅ Required for completion
Internship/job placement support	💼 Linked with regional tech firms
Post-program advancement	🎓 NSF REU, college enrollment, or startup incubation