The attendee’s guide to the AWS re:Invent 2026 AI/ML track

The Attendee’s Guide to the AWS re:Invent 2026 AI/ML Track

AWS re:Invent 2026 runs from Nov 30 – Dec 4, offering more than 140 AI/ML sessions across a full stack of services. Whether you’re a newcomer or a seasoned practitioner, this guide shows how to navigate the AI/ML track, understand the session formats, and prioritize the talks that align with your goals.

---

1. Session Types – How to Learn

| Type | Length | Format | Recording | |------|--------|--------|-----------| | Breakout | 60 min | Lecture‑style presentation | Yes (YouTube) | | Chalk Talk | 45 min | Interactive Q&A with a virtual whiteboard | No | | Builder | 60 min | Hands‑on build‑something activity | No | | Workshop | 120 min | Small‑team problem solving with AWS services | No | | Code Talk | 45 min | Live coding demo led by an expert | No | | Lightning Talk | 20 min | Quick demo or customer story | Yes (YouTube) |

If you can attend in person, mix recorded breakouts (great for reference) with live, interactive formats that let you ask questions and get immediate feedback.

---

2. The AWS AI Stack – Where Sessions Fit

1. Global Infrastructure & AI Chips – Trainium and NVIDIA GPUs power large‑scale training and inference.
2. Data Foundation – Services such as S3, RDS, and OpenSearch provide the data pipelines agents need.
3. Models, Training & Inference – Amazon Bedrock (managed foundation models) and Amazon SageMaker (custom model lifecycle).
4. Agent Development & Management – Bedrock AgentCore lets you orchestrate multi‑model agents.
5. Agentic Solutions – Ready‑to‑use agents like AWS Quick, Amazon Connect, and partner solutions on Marketplace.
6. Security, Governance & Cost – Controls span from chip‑level isolation to IAM policies and cost‑optimization tools.

Understanding this stack helps you map sessions to the layer you’re most interested in.

---

3. Curated Session Paths

A. New to AI on AWS?

• AIM204 – “AI Stack Overview” – Walks through chips, Bedrock, SageMaker, and AgentCore in a single session. Ideal for first‑timers.
• AIM363 – “Governed Data for AI” (Chalk Talk) – Shows how clean, governed data improves model performance and reduces cost.

B. Building & Scaling Models

• AIM364 – “Choosing Between Bedrock & SageMaker” – Real‑world coding agent demo; helps decide when to use managed vs. custom models.
• AIM319 – “Measuring AI ROI” – Live demo of cost‑per‑outcome metrics and business impact analysis.

C. Agent Development

• AIM401 – “AgentCore Deep Dive” – Technical walkthrough of building multi‑model agents, including prompt caching and distillation.
• AIM415 – “Production‑Ready Agents with AWS Quick” – Builder session that connects your tools and data into a personal knowledge graph.

D. Advanced Topics & Optimization

• AIM452 – “Trainium Performance Tuning” – Workshop on maximizing throughput and cost efficiency on AWS‑owned chips.
• AIM470 – “Secure Agent Deployments” – Chalk talk covering IAM, VPC‑isolated endpoints, and audit logging for agents.

E. Sustainability & Infrastructure

• AIM502 – “Green AI at Scale” – Explores how AWS’s carbon‑aware regions and efficient hardware reduce the environmental footprint of large models.

---

4. Planning Your Personal Schedule

1. Identify your priority layer – Chip performance, model lifecycle, or agent orchestration?
2. Select a mix of formats – Pair a recorded breakout (for reference) with a live workshop or code talk for hands‑on experience.
3. Allocate buffer time – Re:Invent’s expo floor, networking lounges, and the re:Play entertainment area are valuable for informal learning.
4. Leverage on‑demand content – All breakouts are posted to the AWS Events YouTube channel within days; schedule catch‑up sessions if you miss anything live.

---

5. Practical Takeaways

• Start broad, then specialize – Begin with an overview session (AIM204) before diving into deep‑technical workshops.
• Focus on cost‑aware practices early – Sessions like AIM363 and AIM319 teach data hygiene and ROI measurement, which pay dividends as you scale.
• Use the AI stack diagram – Keep the six‑layer model handy; it’s a quick reference when scanning the agenda.
• Don’t forget the community – Join the Builder Center Slack channel and the re:Invent Discord for real‑time Q&A with speakers.

By aligning your interests with the stack layers and mixing session types, you’ll leave re:Invent with actionable knowledge, a clearer architecture roadmap, and connections to the AWS AI community.