System Design for AI Agents with Harsha P

This system design course covers the core architecture, data pipelines, and infrastructure required to design, build, and scale production-grade AI Agents.

1. Agent Architecture & Core Logic

• Reasoning Models: ReAct, Chain-of-Thought and Plan-and-Solve patterns.
• Memory Systems: Short-term conversational state vs. long-term vector retention.
• Tool Orchestration: Function calling registries, dynamic API routing and fallback loops.

2. Data Infrastructure & State Management

• Vector Databases: Scaling embedding pipelines, hybrid search and chunking strategies.
• State Machines: Distributed session tracking, checkpoints and multi-agent synchronization.
• Knowledge Graphs: Integrating structured graph data for deterministic context retrieval.

3. Deployment & Scalability

• LLM Serving: Maximizing throughput using batching, caching and open-source engines (e.g., vLLM).
• Event-Driven Runtimes: Processing long-running background tasks via message queues.
• Streaming APIs: Real-time response generation using WebSockets and Server-Sent Events (SSE).

4. Guardrails & Evaluation

• Security: Prompt injection defenses, PII masking and input/output sanitization.
• Agent Evaluation: Automated testing frameworks for execution drift and factual accuracy.

5. Observability & Lifecycle

• Trace Logging: Tracking multi-step tool execution graphs and token latency.
• Cost Optimization: LLM budget management through semantic caching and smart model routing.

Don't miss out on this opportunity to enhance your skills and expand your understanding of system design for AI Agents.

Embrace the journey of learning and explore how you can make an impact in the tech world with System Design today!