Amazon Dev News • Vega Guides & Netflix AI Discover • Nov 9, 2026

Amazon released Vega 2.0 for Fire TV, bringing React Native 0.83 to Vega OS and shipping breaking changes only in annual major releases so app teams can plan upgrades. The company published Hello World and OS compatibility guides plus videos on the update, and announced bots in Kiro Crew alongside a Kiro blog post explaining how Kiro Auto uses model cascading to match models to tasks. Amazon's Kiroween 2026 hackathon runs online through December 4 with over $100,000 in prizes.

Hey team With the Vega 2.0 launch last week, we have a bunch of new docs and videos for you, including guides to Hello World https://developer.amazon.com/docs/vega/latest/media-player-hello-world.html and OS compatibility https://developer.amazon.com/docs/vega/latest/upgrade-os-version-compatibility.html , and videos on What's New in Vega 2.0 https://www.youtube.com/watch?v=1Z12ROcWov8 and React Native 0.83 on Vega OS https://www.youtube.com/watch?v=1zHd71ioj5g . If you use Kiro, we have some guides and videos this week, and we also announced bots in Kiro Crew https://kiro.dev/blog/crew-bots . We also have a bunch of exec interviews with AWS CEO Matt Garman https://www.youtube.com/watch?v=z4sCWxVyx8E , SVP of Prime Video and Amazon MGM Studios Mike Hopkins https://www.youtube.com/watch?v=q9CJnDr115s , Zoox CEO Aicha Evans https://www.youtube.com/watch?v=7J8mpK7GzqM , Ring founder Jamie Siminoff https://www.youtube.com/watch?v=sDh7gMQFnMg , and SVP of Amazon Devices and Services Panos Panay https://www.youtube.com/watch?v=HXTwwznghOw . The Kiroween 2026 https://kiroween.devpost.com/ hackathon is open now and runs online through December 4, and Catherine Cao created a video intro for it https://www.youtube.com/watch?v=PO3M2TKCUb0 . Web Summit https://websummit.com/ starts today in Lisbon, and the AI Engineer Code Summit https://www.ai.engineer/ is tomorrow in San Francisco. This season we're heading to AWS re:Invent 2026 https://aws.amazon.com/events/reinvent/ in Las Vegas, where my colleague Kourtney https://kourtneymeiss.dev/ is leading a workshop, "Turn a Bee wearable into your re:Invent notetaking companion," and if you're building from home, Kiroween 2026 https://kiroween.devpost.com/ is worth a look. We hope to see you there | Date | Event | Location | Description | |---|---|---|---| | Oct 30 – Dec 4 | Kiroween 2026 https://kiroween.devpost.com/ | Online | Kiro's annual Halloween-themed hackathon: build a new app with Kiro in one of four categories for a share of $100,000+ in prizes, with submissions due December 4. | | Nov 9–12 | Web Summit https://websummit.com/ | Lisbon, PT | One of Europe's largest tech conferences, with broad coverage across AI, devices, startups, and platforms. | | Nov 10–12 | AI Engineer Code Summit https://www.ai.engineer/ | San Francisco, US | The code-focused summit in the AI Engineer conference series, on agent and LLM development. | | Nov 30 – Dec 4 | AWS re:Invent 2026 https://aws.amazon.com/events/reinvent/ | Las Vegas, US | AWS's flagship annual conference, with 1,000+ technical sessions, keynotes, workshops, and the largest cloud-community gathering of the year. Dev Advocate Kourtney Meiss https://kourtneymeiss.dev/ is leading a workshop, "Turn a Bee wearable into your re:Invent notetaking companion." | Vega 2.0: What's New for Fire TV Developers https://www.youtube.com/watch?v=1Z12ROcWov8 leads the videos this week, and React Native 0.83 Is Now on Vega OS https://www.youtube.com/watch?v=1zHd71ioj5g covers the framework upgrade that comes with it. For agent builders, How Kiro Auto uses model cascading to match the model to the work https://kiro.dev/blog/model-cascade explains how Kiro picks a model for each task. You'll also find pieces on streaming AI responses, Kiro credit pooling, and conversations with Panos Panay https://www.linkedin.com/in/panos-panay-22b60ba9/ and the CEO of Zoox. Vega 2.0 is generally available and already on every Fire TV device powered by Vega. Anisha Malde https://www.anisha.dev/ , Sr. Developer Advocate at Amazon, breaks down what the move to React Native 0.83 means for your app's speed and your debugging workflow. She also demos how Amazon Devices Builder Tools lets your AI coding assistant upgrade an existing Vega app or bring a Fire OS app over to Vega. Breaking changes now ship only in major releases, once a year, so Vega app teams can plan upgrades around them. This builder.aws.com guide gives your Kiro Crew a remote outpost: a cloud server that keeps agents working on your tasks around the clock while you command it from your laptop like any other crew. It wraps the setup in a starship-fleet metaphor, but underneath is a practical, security-first build with no open ports and no SSH keys to leak, plus real numbers on what keeping a crew on standby costs. It pairs with this week's bots in Kiro Crew https://kiro.dev/blog/crew-bots announcement, for agents that shouldn't stall the moment you close the lid. Vega OS apps built on React Native 0.72 can now move to 0.83. In this talk from NextApp DevCon in Berlin, Steven Moyes https://www.linkedin.com/in/stmoy/ , Sr. Product Manager at Amazon, explains what the upgrade means for Fire TV apps, including better debugging, and how Amazon Devices Builder Tools lets your AI agent plan the migration for you. For written guidance to go with it, see React Native 0.83 for Vega OS https://developer.amazon.com/apps-and-games/blogs/2026/10/react-native-for-vega-os on our blog. By Vijay Lingam https://www.linkedin.com/in/vijaylingam/ , Mayank Gupta https://scholar.google.com/citations?user=vE4HzH8AAAAJ , Sudipto Nandi https://www.linkedin.com/in/sudipto-nandi-2a32175/ , Neil Kulkarni https://www.linkedin.com/in/neilkulkarni1/ , Jack Wang https://www.linkedin.com/in/hanqiw , and Sai Srinivas Somarouthu https://www.linkedin.com/in/saisrinivas789/ The Kiro team explains what actually happens when you leave a session on Auto: each session starts on one model, a background check reviews the work, and the session moves to a stronger model only when the task needs it. It's a short, clear read on why Auto can stay close to top-model quality without running every task on the most expensive model. The internal evaluation results are the part to read when deciding whether to pin a model or trust Auto. At next app devCon 2026 in Berlin, a panel of engineering leaders, Seth Webster https://www.linkedin.com/in/swebster/ , Nicola Corti https://www.linkedin.com/in/cortinico/ , Kadi Kraman https://www.linkedin.com/in/kadi-kraman/ , Ghasem Bakhtiari https://www.linkedin.com/in/ghasem-bakhtiari-/ , Michał Chudziak https://www.linkedin.com/in/michalchudziak/ , Steven Moyes https://www.linkedin.com/in/stmoy/ and Anisha Malde https://www.anisha.dev/ , compares how React Native first came into their companies. Some rewrote their apps from scratch, some added it to existing apps, and some migrated over from other cross-platform frameworks. The panel digs into the production trade-offs at millions of users and what each team would do differently today. Their answers double as a candid reference for anyone weighing React Native for Vega on Fire TV. By Shweta V A https://www.linkedin.com/in/shwetavinodkumar/ , Ranjith Ramakrishnan https://www.linkedin.com/in/ranjithramakrishnan/ , and Krishna Parab https://www.linkedin.com/in/krishnabparab/ The Kiro team explains how credit pooling changes the way teams pay for agentic engineering work. Instead of each seat holding its own allowance, the whole team draws from one shared balance, and admins get controls over who uses what. Team leads get a clear picture of how spend stays predictable when a few heavy users and many light users share one budget. Most AI agents spend their lives waiting on a slow tool, a human reply, or an alert that might not arrive for weeks, and they keep a machine running the whole time. Clare Liguori https://www.linkedin.com/in/clareliguori/ , Senior Principal Engineer at AWS working on Kiro and a core maintainer of the Model Context Protocol, makes the case for reactive agents that sleep as plain data and wake only when there's work to do. She draws on Infinity, one of our experimental projects, then lays out what MCP already supports for this pattern and where the working group still wants feedback. It's a short, practical way to stop paying for idle time. By Ramses Mata https://www.linkedin.com/in/ramses-mata/ On the AWS Developers channel, Ramses Mata https://www.linkedin.com/in/ramses-mata/ explains why AI apps show their answers word by word, and what is actually passing between the model and the client while that happens. The video walks through streaming to the console and then to a browser. It also covers when streaming is not worth it, since errors and guardrails get harder to handle once tokens are already on screen. A companion builder.aws.com walkthrough of streaming in Python https://builder.aws.com/content/3L6Yy7qZJ2PTKTAdPv7psgVzXq2 goes further into why model output arrives in pieces and how those pieces reach your app over the network, then ends with a working Strands Agents build that keeps an assistant in a Fire TV, Alexa, or Vega app feeling responsive instead of stalled. Kiroween 2026 is underway, and in this kickoff on the Kiro channel, Catherine Cao https://www.linkedin.com/in/catherine-ying-cao-26aa207a/ shows how Kiro's AI agents take a project from planning through implementation, testing and debugging. The video has a chapter for each place you can work with Kiro, from the IDE to the command line, so you can skip straight to the one you use and spend your time building your Kiroween project. By Yaroslav Kharkov https://www.linkedin.com/in/yaroslav-kharkov/ , Ferhat Erata https://www.linkedin.com/in/ferhaterata/ , Behrooz Omidvar-Tehrani https://www.linkedin.com/in/behroozomidvar/ , and Rajdeep Mukherjee https://www.linkedin.com/in/mukherjeerajdeep The Kiro team pointed formal verification at its own agentic harness. Kiro models parts of its codebase in Lean, proves theorems about those models, and checks each model against the real TypeScript, Rust, and Python code. The useful part is what happened when a property failed in the model: the team could reproduce the same bug in the shipping code. It's a working example of proofs catching hidden bugs in production code, well outside academic projects. By Karthik Raman https://www.linkedin.com/in/karthik-raman-247b36104/ , Brian Celenza https://www.linkedin.com/in/bcelenza/ Karthik Raman https://www.linkedin.com/in/karthik-raman-247b36104/ digs into how GitHub rebuilt its Git storage so thousands of agents can push to a single repository at the same time. Git was built for people taking turns, and agentic coding breaks that assumption. Teams running coding agents in parallel will see what happens on the server when they all push at once, and which design choices let it hold up. AWS CEO Matt Garman https://www.linkedin.com/in/mattgarman/ joins Ed Ludlow https://www.linkedin.com/in/edward-ludlow-5a7b4280/ on Bloomberg Tech to lay out how AWS is funding its AI infrastructure buildout responsibly while customer demand keeps climbing. It's a quick, high-level read on AWS's growth outlook, and a useful one if your apps run on AWS AI services and you want to know where the investment behind them is heading. Mike Hopkins https://www.linkedin.com/in/mike-hopkins-543a214/ , head of Amazon MGM Studios, joins WSJ Events to unpack the strategy behind Prime Video's ad-supported tier, the MGM acquisition, and our push into live sports. He also touches on how generative AI tools are finding their way into our production work. For streaming and ad-supported Fire TV apps, it's context on where the biggest content bets on the platform are headed. Zoox CEO Aicha Evans https://www.linkedin.com/in/aicha-evans-4365731 sits down with WSJ Events to explain why Zoox builds the whole stack itself, from a purpose-built vehicle with no driver's seat to the ride-hailing service that runs it. It's a short look at how she thinks about scaling robotaxis and where she wants the service to go next. Akamai co-founder Randall Kaplan https://www.linkedin.com/in/randall-kaplan-57992a9/ sits down with Ring founder Jamie Siminoff https://www.linkedin.com/in/jamiesiminoff/ on In Search of Excellence to trace the long road before Ring, from his first campus businesses to a run of investor rejections that nearly ended an earlier company. In part one, Jamie digs into the questions every early-stage builder faces: whether you need a co-founder, how to earn a meeting with investors, and why raising too much can hurt you. His account of being straight with creditors while the business struggled is worth the watch on its own. Erin Lawrence https://www.linkedin.com/in/erin-lawrence-13ba7629/ of Tech Gadgets Canada sits down with Panos Panay https://www.linkedin.com/in/panos-panay-22b60ba9/ , who leads our Devices & Services organization, to talk about the latest hardware, the work he's proudest of and where the team is heading next. It's a look at the thinking behind the devices Fire TV, Alexa and tablet apps run on. Questions or comments? Have an item for next week's post? Let me know https://community.amazondeveloper.com/new-message?username=Moses