The AWS Builder Center is now available as a native mobile app for iOS and Android, and the same week introduced in‑app polls and the Zero to Shipped hackathon; at the same time Amazon Connect Talent, Amazon Corretto 27, and Moonshot AI’s Kimi K3 model on Amazon Bedrock reached general availability.
AWS Builder Center mobile app and new community tools
The mobile client lets users authenticate with their Builder ID and retain session state across devices. From a phone you can browse trending articles, launch any of the 600+ Skill Builder courses, and start hands‑on workshops that provision free sandbox environments. The app also surfaces Builder Loft calendars, push notifications for subscribed topics, and a Wishlist button that forwards product feedback directly to AWS product teams. It is distributed worldwide through the Apple App Store and Google Play Store.
Two community‑focused features were added. Polls let any user post a question on the Home feed, define two to five answer choices, set a deadline, and collect anonymous votes; results update live and discussion continues in the comment thread. Only aggregate percentages are shown to the poll creator. The Zero to Shipped hackathon runs from 18 September to 2 October, inviting participants to connect a coding agent to AWS, build a production‑ready application, and deploy it live for a chance at a share of a $28 000 prize pool. Five projects receive $5 000 in AWS credits plus a swag bundle.
Amazon Connect Talent reaches GA
Amazon Connect Talent, an AI‑driven hiring solution, is now generally available. The service uses AI agents to conduct structured voice interviews, administer evidence‑based assessments, and generate competency scores tied to interview evidence. Candidates can interview 24/7 from any device, while recruiters receive transcripts, scores, and detailed evaluations the next morning. All candidate data is anonymized during AI evaluation, and recruiters retain final hiring authority. GA includes competency‑based assessments, adaptive voice questioning, a brandable mobile‑first candidate portal, and admin onboarding tools.
Amazon Corretto 27 GA adds modern JVM defaults
Amazon Corretto 27, the latest feature‑release of the free OpenJDK distribution, is now generally available for Linux, Windows, and macOS with support through April 2027. Key runtime changes are: G1 becomes the default garbage collector (JEP 523); TLS 1.3 uses a post‑quantum hybrid key exchange (JEP 527); object headers are compacted by default to reduce memory footprint (JEP 534); and Java Flight Recorder now redacts sensitive data before it leaves the JVM (JEP 536). The release continues to preview enhanced pattern matching, structured concurrency, lazy constants, and the incubating Vector API.
Kimi K3 on Amazon Bedrock expands open‑weight model options
Moonshot AI’s Kimi K3 model is now generally available on Amazon Bedrock. At 2.8 trillion parameters it is the largest open‑weight model on the service and adds native vision capabilities plus a 1 million‑token context window, making it suitable for long‑running code analysis, multi‑document queries, and extended agent workflows. Moonshot AI reports roughly 2.5× better scaling efficiency versus its Kimi K2 predecessor. Kimi K3 also supports explicit prompt caching, which can lower latency and input cost when reusing context across calls.
Related CloudNinjas coverage: AWS.
What This Means For Practitioners
Mobile access to Builder Center means engineers can consume training and spin up sandbox workshops without a laptop, enabling on‑the‑go learning and rapid prototyping. Polls provide a lightweight feedback loop for community‑driven decisions, while the Zero to Shipped hackathon offers a structured path to production deployments and credit incentives.
Connect Talent introduces an AI‑augmented interview pipeline that can reduce manual screening effort, but teams must still validate AI‑generated scores and ensure anonymization policies align with internal compliance.
Corretto 27’s default G1 collector and post‑quantum TLS settings shift baseline performance and security expectations; existing Java workloads should be tested for GC pause behavior and TLS compatibility before upgrading.
Kimi K3’s large context window and prompt caching open new design patterns for code‑intensive agents on Bedrock, but the model’s size and cost profile should be evaluated against workload requirements.
Overall, the announcements tighten the feedback loop between learning, building, and deploying on AWS, while the GA releases raise the baseline capabilities of AI hiring, Java runtimes, and open‑weight foundation models.
