Architecting Flexible Agentic AI on Amazon EKS, with NVIDIA and LangChain (Singapore)
👉 Registration is open. Please visit the AWS landing page. 👈
Workshop Overview
Agentic AI is moving into workloads where the data cannot leave the building — regulated industries, government, healthcare, and any organisation whose most valuable data is also its most restricted. This workshop covers how to build agentic AI on infrastructure you control: the architecture for self-hosting open-weight models on Amazon EKS, the open-model landscape and where it is heading, and a hands-on build session with LangChain where you construct a working agent yourself. You'll leave with an agent you built, a deployment pattern you can reproduce, and a clear view of the choices available on where your intelligence runs.
If you're asking any of these questions, this workshop is for you:
"My most valuable data can't be sent to a third-party API — how do I still build agents on it?"
"How do I run open-weight models inside my own environment without building a serving platform from scratch?"
"How do I keep the same agent portable across in-region cloud, on-premises and disconnected environments?"
"How do I own the model — the weights, the fine-tune, the behaviour — rather than rent it?"
"How do I know my agent got better, not just different, when I change a prompt, a tool or a model?"
What You Will Build (hands-on, with LangChain)
A working agent using LangChain — orchestration, tool calling, state and control flow
The agent structured so the model is a swappable component, not a vendor dependency: managed endpoint, open weights or your own fine-tune, without rewriting the application
An evaluation loop, so you can tell whether a change to a prompt, a tool or a model actually improved the agent
The whole thing running on Amazon EKS, against models served inside the workshop environment
What You Will Learn (sessions and demos)
Sovereign AI architecture on Amazon EKS – how to keep weights, prompts, inference and telemetry inside a boundary you define, with Kubernetes as the portable substrate across in-region, on-premises and edge
Open models and NVIDIA Nemotron – the open-weight model landscape, what Nemotron is built for, and where open models win on data control, cost, latency and auditability
The agent operating model – how to structure, version, observe and govern agents so they are owned by a team and safe to change
Fine-tuning and harness optimisation – a live demonstration of tuning the agent harness and fine-tuning an open model on private data, and how to measure whether it helped
Running inference on Kubernetes – GPU right-sizing, scheduling and scaling considerations for inference you operate yourself
Who Should Attend
AI/ML Engineers deploying LLMs, agents and RAG pipelines on sensitive or regulated data
Platform Engineers & AI Ops building the sovereign serving and orchestration layer other teams consume
Backend & application engineers integrating agents into products with data-residency obligations
DevOps & SRE teams responsible for AI workload reliability, cost and scale
Engineering leads and architects making open-vs-managed and build-vs-rent decisions on model serving
Public sector, financial services, healthcare and defence teams whose data cannot leave a defined jurisdiction
Bring your own use case. The day closes with 1:1 time with AWS, NVIDIA and LangChain engineers to work through your specific sovereignty and architecture constraints.
Agenda
09:30 – 10:00 — Registration and walk-in
10:00 – 10:15 — Welcome address
10:15 – 10:45 — Architecting flexible, sovereign AI on Amazon EKS (AWS)
10:45 – 11:15 — Open models: owning the weights (NVIDIA)
11:15 – 11:45 — Own your intelligence: the Agent Operating Model (LangChain)
11:45 – 14:00 — Lunch, networking and workshop environment setup
14:00 – 16:00 — Hands-on workshop: build an agent on Amazon EKS with LangChain
16:00 – 16:30 — Fine-tuning and harness optimisation, live demo (LangChain and NVIDIA)
16:30 – 17:00 — 1:1 clinic: bring your own use case, with AWS, NVIDIA and LangChain engineers
Agenda subject to change.
Register Now → here.
Seats are limited. Register now to secure your spot and start building agentic AI you own, on infrastructure you control.
About LangChain:
LangChain enables every company to own their intelligence. LangChain's open, model-agnostic harnesses give teams choice and control over how they build their agent architecture. LangSmith brings testing, deployment, and monitoring together so teams can continuously improve their agents, compound intelligence, and govern them at scale. More than 7,000 customers, including Nvidia, Bridgewater, LinkedIn, Workday, Harvey, and Rippling trust LangSmith to improve their agents across the Agent Development Lifecycle. Learn more: www.langchain.com