The One About Agents & Evals (ft. LangChain & Huawei)
As agentic systems grow more complex, orchestration and evaluation are becoming core engineering challenges. Explore how modern agent frameworks coordinate tools, memory, reasoning, and human oversight across complex workflows, alongside practical approaches for evaluating and improving reliability in production.
More About the Sharings
Yu Tao & Ding Yangkai (Technical Experts, Huawei) will share on "Coordination Engineering"
As multi-agent systems grow increasingly capable, coordinating communication, task delegation, and efficient collaboration among agents has become a critical systemic challenge. Yu Tao & Yangkai will share on "Coordination Engineering", a modeling approach developed using the open-source openJiuwen framework to build collaborative architectures across multi-agent systems.
Drawing on the latest research, explore how to design coordination patterns, task decomposition, communication structures, and execution workflows – with the goal of improving robustness and systematic collaboration among agents within complex workflows. (Technical Level: 100)
Liam Bush (Deployed Engineer, LangChain) will share on “Agent Architectures and Harnesses”
As agentic systems become increasingly stateful and multi-step, orchestration frameworks such as LangGraph are emerging as key building blocks for production workflows. At the same time, Agent Harnesses like, Deep Agent and Claude Agent SDK, have gained popularity. Liam will explore how deep agent architectures are designed in practice, including graph-based orchestration, state management, branching logic, and human-in-the-loop patterns used in real-world applications. (Technical Level: 200)
Allen He (APAC Lead, LangChain) will share on “Agent Evaluations - from prototype to production"
According to MIT, 95% of agents fail to move from prototype to production. Allen will share the best practices from leading enterprises and startups on overcoming the conflict between agent reliability and LLM models' non determinism. From there, he will explore practical approaches to agent evaluation in production environments, including debugging workflows, robustness testing, observability patterns, and evaluation methodologies used to assess agent behaviour across real applications. (Technical Level: 200)
More About the Speakers
Yu Tao is a technical expert at Huawei with over 10 years of experience in artificial intelligence, specialising in agent technologies and natural language processing. His work spans applied AI systems and research, with multiple publications at leading CCF-A/B conferences in areas related to LLMs, agent systems, and NLP.
Ding Yangkai is a technical expert at Huawei focusing on agent technologies and retrieval-augmented generation (RAG) systems. His research explores multi-agent architectures, reasoning workflows, and knowledge-enhanced AI systems, with multiple publications at leading CCF-A/B conferences.
Liam Bush is a Deployed Engineer at LangChain, where he works on helping teams build, evaluate, and deploy production-grade agentic systems. His work spans applied AI engineering, aent implementation, and supporting organisations in moving from prototype workflows to reliable, scalable agentic systems. Prior to LangChain, he was an Applied AI Engineer at Valiant Solutions, building production AI systems across applied NLP and enterprise use cases. He holds a BSE in Computer Science from the University of Pennsylvania, with a focus on artificial intelligence.
Allen He is the APAC lead at LangChain, where he works with enterprises to adopt and scale agentic AI solutions. He focuses on helping organisations move from prototyping to production, and creating lasting organisational changes suited for the new world of agent development. Prior to LangChain, Allen held roles at McKinsey, LinkedIn, Salesforce, and NetApp, where he worked across strategy and transformation. He holds a business degree from UC Berkeley and MBA from Wharton with a focus on entrepreneurship.
More About the Series
AI Wednesdays is Lorong AI’s weekly gathering, bringing together practitioners, researchers and innovators for technical discussions on research insights, product development and engineering practices.
Get involved: Learn more about Lorong AI | Speaker Sign-up | WhatsApp Community | LinkedIn | X