here are 4 screenshots all featuring me. Clearly playing with #sora2 has aged me horribly

If you are at #ICML2025 next week, come stop by the Shopify booth, I'd love to chat with you! If not, take a look at the papers and let me know what's interesting to you! I'll talk about them in upcoming video and will reach out to the authors at the conference if you give me

On Android Go to Settings → Network & Internet → Wi-Fi. Tap your connected network → Advanced or Edit. Look for IP settings → change to Static. Scroll to DNS 1 and DNS 2, and enter new addresses.

On iPhone/iPad Go to Settings → Wi-Fi. Tap the i next to your network. Scroll to DNS → tap Configure DNS → choose Manual. Delete existing servers, then add new ones 8.8.8.8. Tap Save.

On macOS Open System Settings (or System Preferences). Go to Network → choose your current connection. Click Details… (or Advanced…) → go to the DNS tab. Click + and enter new DNS servers like 8.8.8.8, etc. Click OK → Apply.

On Windows 10/11 Press Win + R, type ncpa.cpl, and press Enter. Right-click your active network (e.g., Wi-Fi or Ethernet) → Properties. Select Internet Protocol Version 4 (TCP/IPv4) → Properties. Choose Use the following DNS server addresses, and enter: 8.8.8.8

If #CenturyLinkdown is ruining your day change your DNS server (google's is 8.8.8.8)

About to host a panel at this free event, come join us! https://t.co/dAp994Ha0A #MLOpsCommunity

ChatGP-who? Suddenly, DeepSeek is the new AI on the block. China’s latest model is making waves—not just for its capabilities, but for the engineering techniques behind it: 🔹 Sparse Mixture of Experts 🔹 Multi-head Latent Attention 🔹 Model Distillation I’m excited to chat

Got pushback on my last video—some claim seniority is all that matters in ML hiring. That’s 100% garbage. Less senior candidates get hired all the time. What actually gets you interviews: ➜ Pedigree (Stanford + Bengio’s lab + FAANG? You’re in.) ➜ Referrals (Your former

Neural networks are ‘inspired by biology,’ but imagine if as a newborn, you were given 80% of all human knowledge—shuffled, out of order, and never learned anything new again. That’s how we train AI today. Maybe it’s time for a rethink? 🤔 #AI #MachineLearning

🚨 ML demos should build better AI, not just hype. Stop faking progress—learn how to use iterative demos to make real ML advancements. 👇 🔗 https://t.co/V0KZPfKKkC #MachineLearning #AI

Just heard from an M2 candidate I’ve been working with—FAANG company finally got back to them… 3 MONTHS after their interview. Turns out, not all ghosting means no. 👀 #FAANG #InterviewTips

🚀 Want to level up your ML knowledge? Kevin Van Horn (30+ years in ML, ex-Adobe Sr. Staff) dives into Bayesian probability, optimization, & the hard truths of ML. A must-listen for ML engineers & researchers! 🎧👇 https://t.co/HSkU6PxC6E #MachineLearning #Podcast

Most decisions in ML engineering aren’t one-way doors—you can walk back through them. But early in your career, everything feels irreversible. The more experience you gain, the more you learn to design systems with escape routes. Master two-way doors, and you’ll move faster. 🚀

#Meta warns that is will fire leakers in a leaked memo Now that's meta. https://t.co/RPGmkFhgcO

Working with a Senior Staff ML candidate in office hours yesterday reminded me: High-level ML System Design interviews aren’t just about covering all stages—you must save time for a deep dive. Here’s my breakdown of how to structure your time. By the way, I host office hours

If you’re serious about preparing for ML System Design Round, book a mock interview here: https://t.co/lTmWMEwbj3 🚀

Many candidates make the mistake of only reviewing generic ML concepts. Instead, go deep into how these systems scale, their trade-offs, and real-world constraints. #MLEngineering #TechInterviews

The most common ML System Design topics at Meta are: ➜ Recommender Systems – Powering content ranking & personalization ➜ Harmful Content Detection – Filtering spam, misinformation & safety risks ➜ Topic Modeling (NLP) – Predicting category of conversation #AI #DataScience