- TechTok Newsletter
- Posts
- Apple New App 🔥, Claude Voice 🗣️, Discord Pays Users 💰
Apple New App 🔥, Claude Voice 🗣️, Discord Pays Users 💰
Today's TechToks: Apple replacing Game Center, Anthropic Claude announces voice, Discord pays for ads interaction, Trending GitHub repositories, Fresh Product Picks, and much more!
TechTok by João Martins
May 28, 2025
See all as stories in techtok.today
Today's Picks🍒
🚀 Apple to debut new app, Claude launches voice, discord pays users for ad interaction
🦾 From Meta Engineering: Scaling Instagram’s recommendation system to 1000+ models
🌟 GitHub's Trending Repositories
📈 Time tracking app
🔀 Serverless Postgres
⚙️ Machine learning engineering agent
🔥 Products of the Week
🙌 A visual workspace for your browser tabs.
🗣️ Translate apps with AI, no proofreading.
🛬 Get customized landing pages for your website.
➕ And much more
🧐 Weekly Picks
Most impactful articles and news of this week on techtok.today
Apple will launch a new gaming app this fall, replacing Game Center with a centralized hub for achievements, social features, and Apple Arcade. Announced days before Nintendo Switch 2’s debut, it signals Apple’s renewed push to become a serious gaming platform.
bloomberg.com
More Tech News
📱 Product Picks
Curated products from Product Hunt.
🔥Trending products of the week
Simple screen time app. The more you brainrot, the more your brain rots. Keep your cute little brain avatar healthy by limiting your screen time.
A visual workspace for your browser tabs. Create columns, group tabs, and set smart reminders (Chrome, Firefox, and Safari).
💼 Bolto
Find, interview, and hire engineers with AI. Backed by Y Combinator, it also enables startups to pay and manage those engineers in the same place.
🗣️ Algebras AI
Translate apps with AI, no proofreading. Translate and localize with an open-source tool and a clean UI, built for experimenting fast.
Get customized landing pages for your website. Each landing page is tailored for your prospect accounts, contacts, industry, persona, ad keywords, and more.
👩🏽💻 Engineering Blogs
Articles from engineering blogs of big tech companies
 |
Scaling ML requires infra-first thinking—tools for visibility, safety, and speed
The Problem
Instagram’s recommendation system ballooned to 1000+ ML models—each powering feeds, stories, comments, and tags. But rapid growth led to:
No centralized tracking (teams lost visibility).
Slow, risky launches (days per model).
Unreliable predictions (degraded recommendations).
The Solution
To tackle these challenges, Instagram’s engineering team built a robust infrastructure framework focused on scalability, automation, and reliability. Key innovations included:
1️⃣ Model Registry – A single source of truth for model metadata, criticality, and ownership.
2️⃣ Automated Launches – Cut launch time from days → hours with pre-tested capacity scaling.
3️⃣ Stability Metrics – Real-time tracking of prediction accuracy (calibration & entropy) to catch failures.
The Results
These implementations brought over 10 weekly launches, with 2-day reduction in launch time, and hidden model issues being discovered and fixed faster— leading to higher-quality recommendations for billions of users.
✨ "Too long don't README"
GitHub trending repos of today
Time tracking application designed for freelancers and agencies. Features include time tracking, project and task management, billable rates, and support for multiple organizations. Allows data imports from other time tracking tools.
What can you use this for? Track time with productivity reports, manage projects and tasks, calculate work-hour costs
Serverless Postgres. With separated storage and compute to offer autoscaling, code-like database branching, and scale to zero. It consists of compute nodes and a scalable storage backend called Pageserver, with Safekeepers providing a redundant WAL service.
What can you use this for? Run small experiments, test code changes, and use serverless, scalable PostgreSQL in your applications.
Repos to learn something new
Machine learning engineering agent to automate the most critical and valuable aspects of the industrial R&D process. With easy exploration of datasets, it LLMs and specialized modules to streamline critical data science workflows.Outperforms other AI systems on the MLE-bench benchmark.
What can you use this for? Automate model development, data preprocessing, and leverage for quantitative trading.
