100 Python Projects: Advanced Agentic Systems 🚀
A Comprehensive Journey from Scripting to Autonomous Swarms.
📖 Overview
This repository contains 100 self-contained Python projects designed to master the field of AI Engineering and Agentic Workflows. The curriculum progresses from basic structured outputs to complex multi-agent simulations, safety evaluations, and governance frameworks.
Status: ✅ Completed (100/100 Projects)
🏗️ Tier 1: Agent Foundations (Projects 1-30)
Focus: The building blocks of LLM interactions and tool use.
- Prompt Engineering: Templates, Chains, Summarizers (Projects 1-10).
- Tool Use: Calculators, Search, File I/O, API Fetchers (Projects 11-20).
- Memory & State: Chat History, Vector Stores, Context Management (Projects 21-30).
🧩 Tier 2: Multi-Agent Architectures (Projects 31-60)
Focus: Orchestration, collaboration, and complex system design.
- Communication: Message Buses, Pub/Sub, Broadcasting (Projects 31-40).
- Patterns: Master-Worker, Hierarchical Teams, Voting, Bidding (Projects 41-50).
- Simulations: Economies, Swarms (Boids), Ecosystems, Debates (Projects 51-60).
🛡️ Tier 3: Evaluation, Safety & Ethics (Projects 61-100)
Focus: Making agents robust, safe, compliant, and trustworthy.
🔍 Assessment & Metrics (61-75)
- Unit Testing: Mock LLMs, deterministic evals.
- Observability: Tracing, Logging, Dashboards (OpenTelemetry, Rich).
- Optimization: Caching, Cost Tracking, Performance Profiling.
🔐 Security & Adversarial Testing (76-80)
- Red Teaming: Prompt Injection, Jailbreaking, Adversarial attacks.
- Defenses: PII Redaction, Input Sanitization, Access Control (RBAC).
⚖️ Safety & Alignment (81-85)
- Constitutional AI: Self-correction based on principles.
- Bias Detection: Demographic parity testing.
- Hallucination Prevention: Consistency checks and citation verification.
🏛️ Ethics & Governance (86-95)
- Documentation: Model Cards, Data Lineage.
- Compliance: GDPR Checking, License Compliance.
- Forensics: Audit Log Analysis, Impact Assessments.
🧑🤝🧑 HCI & Integration (96-100)
- Collaboration: Human-in-the-Loop (HITL), Collaborative State.
- Orchestration: Intent Routing, Notification Management.
- The Omnibus Agent: A final system integrating all patterns.
🚀 How to Run
Each project is self-contained in its own directory with a src/main.py and README.md.
# Example: Running the Final Omnibus Agent
python Tier-3-Evaluation-and-Safety/100-Final-Omnibus-Agent/src/main.py
🛠️ Technologies Used
- Languages: Python 3.10+
- Standard Lib:
threading, queue, json, dataclasses, sqlite3, vcr
- Simulation:
mock, unittest, Custom Event Loops
- Visualization:
rich, matplotlib (simulated), mermaid
Built with ❤️ by Antigravity during the “100 Projects” Challenge.