Vincent Ong

@thsottiaux People ask me what do you do when limits are maxed for Codex, I say just look for Tibo 😂

Codex usage limits have now been reset across all paid plans. Enjoy the weekend!

Life’s short, why be afraid? Chase your dreams.

Don’t worry about failing. At least you tried.

@sama @kr0der 5.5.5.55am/pm?

@kr0der i personally think worth it yeah maybe we will reset limits soon

should i try this? i’ll need a 2nd $200 subscription to make it sustainable

@thsottiaux Work on Codex app feels slightly slower vs Codex CLI on a terminal.

What are we obviously not getting right with Codex?

Very productive weekend. Looking forward to something magical on 5.5.5.55pm. @thsottiaux

emergency codex hackathon on a sunday with 20 demos!! everyone cooked 🔥 favorite projects - sticker generator featuring none other than @gabrielchua (by @brianchew ) - gatchapon character generator by @sentrytoast - memory graph and therapy agent for couples - searching up singapore parliament by @rachpradhan thanks @gabrielchua and @brianchew for hosting!

Let’s go!

Have been procrastinating about using the Codex app since GPT 5.5 came out because I’m so used to working with tmux. Started using it an hour ago because I’m attending the Codex Hackathon hosted by @gabrielchua tomorrow. I must say, it’s a refreshing experience. Amazing job to the team at @OpenAI! 👏🏻

@betomoedano This looks good! What are you using to record the video?

Yesterday I did some code maintenance on my AI tattoo app. I made some changes that required looking at the entire codebase and refactoring piece by piece. For instance, I migrated to: - react-native-ease - react-native-pulsar - native colors - removed a bunch of dead code I’m surprised to say that all that work only took one afternoon. I didn’t write anything by hand. I just used the agent skills provided by each package. I don’t even know how to create a simple haptic with Pulsar because I didn’t really read the docs lol. I did tweak some values manually to get exactly what I wanted tho. Maybe that’s the new normal. Do I really need to memorize every library I use? The app definitely feels faster and smoother. Wild times.

Thanks for reviewing the PR! @krystofwoldrich

Just realised my contribution got merged for Expo a month ago. Can't wait to contribute more! @expo

Running reports on the same Aurora instance as your app? Add an Aurora Replica. Replicas share the cluster storage volume, so there's no replication lag for writes and no ETL setup. Reports hit the replica, your app stays fast. Cheapest fix for ReadIOPS and CPU contention.

FSx for Lustre is the AWS HPC file system for Linux. It links natively to S3, so hundreds of EC2 instances share a high-speed parallel file system during compute, then output flows back to S3 for long-term storage. One architecture covers both performance and persistence.

AWS PrivateLink lets you expose a service to another VPC without opening your entire network. Provider creates a VPC Endpoint Service (NLB-backed). Consumer creates an Interface Endpoint. Traffic is private, one-directional, and scoped to that single service only.

Slow API for global users? Don't reach for CloudFront. Global Accelerator routes users to the nearest AWS edge, then over AWS's private backbone to your region. CloudFront caches static content. Global Accelerator speeds up dynamic API calls. Different tools, different problems.

To provide the best front end UX, have low latency, route traffic to nearest edge location and provide static IP addresses for entry into the application endpoints, configure Global Accelerator to forward requests to a NLB and use EC2 instances for the application in an ASG.

To migrate on premises data Center to AWS and use only a certain Region and admins are not permitted to connect VPCs to the internet, use Control Tower to implement data residency guardrails and use Organisations to configure service control policies (SCP).

For tiered applications consisting of EC2 instances decoupled with SQS and DynamoDB storage layer, use EC2 auto scaling target tracking policy to scale out processing tier instances. Use ApproximateNumberOfMessages attribute to determine when to scale.