Agentic Systems - AgentCamp

A comprehensive, AgentCamp designed to take participants from foundational AI concepts to advanced agentic systems, evaluation frameworks, and real-world architectures. This AgentCamp blends theory with practical design patterns, focusing on building robust, scalable, and production-ready AI agents.

1. Introduction to Agentic AI

Description

Evolution of Artificial Intelligence, with a focus on the emergence of Agentic AI systems. Participants will understand how modern AI systems transition from passive models to autonomous, goal-driven agents capable of planning, reasoning, and acting.

Topics

• Evolution of AI → ML → Deep Learning → Generative AI
• What are AI Agents?
• Agent vs Workflow vs Automation
• Core components: LLM, Tools, Memory, Planning
• Agent lifecycle (perception → reasoning → action)
• Single-agent vs multi-agent systems
• Real-world use cases and architectures

2. Entire Ecosystem of Agentic AI

Description

A deep dive into the Agentic AI ecosystem, covering frameworks, infrastructure, orchestration layers, and developer tooling required to build and scale AI agents in production environments.

Topics

• Agent frameworks (LangChain, Google ADK, etc.)
• Model providers and APIs
• Tooling ecosystem (vector DBs, retrievers, APIs)
• Orchestration layers and agent runtimes
• Observability and debugging tools
• Deployment architectures (local, cloud, hybrid)
• Cost, latency, and scalability trade-offs

3. LLM as a Judge

Description

Learn how to leverage LLMs themselves as evaluators for outputs, enabling scalable, automated evaluation pipelines. This session focuses on LLM-based scoring, critique generation, and alignment validation.

Topics

• Why evaluation is hard in generative systems
• LLM-as-a-judge paradigm
• Prompting strategies for evaluation
• Pairwise comparison vs rubric-based scoring
• Bias, variance, and calibration issues
• Human-in-the-loop vs automated evaluation
• Use cases: ranking, filtering, safety validation

4. Agent Memory

Description

Explore how agents retain and utilize memory across interactions. This session covers short-term, long-term, and episodic memory systems, and how they influence reasoning and personalization.

Topics

• Types of memory: working, episodic, semantic
• Memory architectures in agents
• Vector databases and retrieval mechanisms
• Memory compression and summarization
• Context window management
• Personalization and user modeling
• Memory consistency and drift challenges

5. Agent Simulation & Evaluation

Description

Understand how to test and validate agent behavior using simulation environments and evaluation frameworks. Learn to design robust evaluation pipelines for agent reliability and performance.

Topics

• Simulation environments for agents
• Synthetic data generation
• Scenario-based testing
• Metrics: success rate, latency, cost, accuracy
• Offline vs online evaluation
• Failure modes and robustness testing
• Continuous evaluation pipelines (CI/CD for agents)

6. RAG, Graph RAG, Agentic RAG

Description

A comprehensive exploration of Retrieval-Augmented Generation (RAG) and its evolution into Graph RAG and Agentic RAG systems for complex reasoning and knowledge retrieval.

Topics

• Fundamentals of RAG
• Embeddings and similarity search
• Chunking and indexing strategies
• Graph RAG: knowledge graphs + retrieval
• Multi-hop reasoning
• Agentic RAG: tool-augmented retrieval loops
• Hybrid search (keyword + semantic)
• Performance optimization and pitfalls

7. AI Protocols (MCP, A2A)

Description

Emerging AI communication protocols that enable interoperability between agents, tools, and systems. Focus on how protocols standardize interactions and enable scalable ecosystems.

Topics

• What are AI protocols?
• Model Context Protocol (MCP)
• Agent-to-Agent (A2A) communication
• Tool interoperability standards
• API vs protocol-based communication
• Security and access control
• Future of decentralized agent ecosystems

About the Speaker

Rama Krishna Raju Samantapudi is a seasoned AI/ML Architect and Scientist with deep expertise in building scalable AI/ML systems, LLM-powered applications, and agentic workflows. The specializes specializes in translating cutting-edge research in Generative AI and autonomous agents into production-grade solutions that drive real-world impact.

With extensive experience across Search, NLP, Conversational AI, Agentic AI, and large-scale AI workflows, the speaker has led the design and development of end-to-end intelligent systems.

His work focuses on:

• Designing agentic AI architectures for enterprise use cases
• Building LLM-powered platforms with strong evaluation and guardrails
• Developing scalable retrieval and knowledge systems (RAG, Graph RAG)
• Advancing Responsible AI and governance frameworks
• Enabling organizations to operationalize AI at scale

Rama is also passionate about sharing knowledge to help engineers and leaders adopt modern AI paradigms effectively.

🏢 About the Organizing Committee

Global AI Community – Austin Chapter! We’re a passionate group of AI builders and learners exploring AI, Copilot, Agents, automation, and the future of intelligent technologies. Our mission is to learn, share, and innovate together. Through meetups, workshops, and hands-on sessions, we explore modern AI tools, real-world use cases, and emerging trends shaping the future of work and development. Whether you're beginning your AI journey or already building advanced solutions, the Austin Chapter is your space to connect, grow, and collaborate. We look forward to seeing you at our upcoming events

🎯 Mission

To create a high-impact learning ecosystem that bridges the gap between AI theory and real-world implementation, empowering participants to design, build, and scale intelligent systems.

🚀 What We Do

• Curate industry-relevant workshop content aligned with latest AI trends
• Facilitate hands-on learning experiences and live demos
• Enable collaboration between practitioners, researchers, and leaders
• Promote best practices in building reliable and responsible AI systems

🤝 Community Focus

The committee is committed to building a strong AI community, encouraging:

• Open knowledge sharing
• Cross-disciplinary collaboration
• Continuous learning and experimentation

Together, the speaker and organizing committee aim to deliver a AgentCamp experience that is practical, forward-looking, and deeply technical—equipping participants to lead in the era of Agentic AI.