Dan Huang

You are in a system design interview at Meta for an IC4 role. The interviewer leans in and asks: "If a photo is already on my phone, how can WhatsApp still delete it when the sender taps Delete for Everyone?" Here is how you should break it down 👇 A lot of people think “Delete for Everyone” means WhatsApp somehow reaches into your gallery and erases a file from your device. That is not really what is happening. Btw, if you’re preparing for system design/coding interviews, check out my free mock interview tool on Layrs. You can use it for free here: https://lnkd.in/gpCn7t2T The real answer depends on where the photo is stored, how the chat app references it, and what exactly the app is allowed to delete. 1. First clarify the product behavior Before jumping into architecture, say this clearly in the interview: There are two different cases here. a) The photo exists only inside the chat app’s managed storage b) The photo has already been saved to the user’s gallery / camera roll / filesystem This distinction is everything. If the photo is only inside WhatsApp’s controlled storage, the app can remove the local file and remove the message reference from the chat UI. If the photo was exported or auto-saved to the gallery, WhatsApp usually cannot reliably delete that copy, because that file is now outside the app’s normal ownership boundary. So “Delete for Everyone” is mostly about: - deleting the message record - deleting the media attachment reference - deleting any app-managed cached copy - syncing that deletion event to all participants It is not magic remote file deletion across the whole phone.  2. High-level idea The clean way to think about this is: A chat message is metadata plus optional media. For a photo message, the system usually has: - message_id - chat_id - sender_id - media_id - timestamp - status - optional encryption metadata And the actual photo itself may live in: - encrypted object storage on server for temporary delivery - app sandbox / local database / media cache on device - optionally the user’s gallery if exported When sender taps Delete for Everyone, the system does not chase every byte everywhere. It sends a deletion command tied to the message_id.  3. What happens when the photo is first sent Here is the send flow: 1. User picks a photo 2. App encrypts media and prepares message metadata 3. App uploads encrypted media blob or sends it through media pipeline 4. App sends message metadata referencing that media 5. Receiver downloads the encrypted media 6. Receiver stores it locally in app-managed storage 7. Chat UI shows the message by resolving message_id -> media_id -> local file Read rest of the post: https://lnkd.in/gpv-wGze

158

12 Comments

https://lnkd.in/g8Qu85Yj Adam Walker looks at how AI is forcing us to rethink writing and education. “Hu's recent article in the New Yorker asks, "What happens after AI destroys college writing?" …According to a survey conducted by the digital education council, 86% of students use AI to complete their studies at university…. …The real problem is not AI. It's actually something much deeper than that. For the past 60 years, higher education has modeled itself around efficiency and customer satisfaction. Universities now must recover the value of slow thinking, recursive writing, and the art of communication, delivery, and persuasion. And I believe that universities should scrap college composition and return to the teaching of rhetoric, the art of communication, and persuasion…   Composition teaches students how to write a five paragraph essay. But rhetoric prepares students to write college papers and how to compose a eulogy for their grandfather. How to write a best man speech at a wedding or a made of honor speech. Teaches them to address their local school board and to persuade a group of shareholders at a corporate job. And it teaches them to address citizens of a nation... Rhetoric is not the mastery of technical skill alone but the formation of the entire imagination. And this I think is partly what a return to rhetoric would offer students.”

1

🚀 Fine-Tuning Small Language Models (SLMs): A Step-by-Step Guide As AI adoption accelerates across industries, Small Language Models (SLMs) are emerging as powerful, cost-effective tools for domain-specific tasks — especially in regulated sectors like healthcare. Here’s a practical guide to fine-tuning SLMs for your enterprise use case: 🧠 Step 1: Define the Use Case Start with a clear objective — e.g., member journey optimization, clinical note summarization, or marketing content classification. 📦 Step 2: Prepare the Dataset Curate high-quality, domain-specific data. Clean and tokenize text. Ensure compliance with data privacy (HIPAA, etc.). 🏗 Step 3: Choose the Right SLM Options: TinyLlama, DistilBERT, BGE-Small, etc. Consider model size, latency, and inference cost. 🛠 Step 4: Set Up the Environment Use Databricks, Hugging Face, or Azure ML. Leverage GPU clusters for faster training. 🔁 Step 5: Fine-Tune the Model Use transfer learning with domain-specific data. Apply techniques like LoRA or QLoRA for efficiency. Monitor training loss and validation accuracy. 🔍 Step 6: Evaluate & Test Use benchmark datasets and real-world samples. Measure precision, recall, F1-score. Perform bias and safety checks. 🚢 Step 7: Deploy Securely Containerize with Docker. Use Azure endpoints or Databricks Model Serving. Implement RBAC and audit logging. 📈 Step 8: Monitor & Improve Track performance in production. Collect feedback and retrain periodically. 💡 SLMs are not just lightweight — they’re agile, interpretable, and ideal for enterprise AI. If you're working in healthcare, MarTech, or regulated domains, now is the time to explore their potential. Let’s connect if you're exploring SLMs for your next AI initiative! #AI #SLM #Databricks #HealthcareAI #MarTech #FineTuning #LLM #AzureML #MLOps #EnterpriseAI #DataScience

14

A new study from Carnegie Mellon, “Does AI-Assisted Coding Deliver? A Difference-in-Differences Study of Cursor’s Impact on Software Projects” that looks at a more recent period, tracking code quality in projects using GenAI tools up to mid-2025, finds that AI briefly accelerates code generation, but the underlying code quality trends continue to move in the wrong direction. https://lnkd.in/gm6dxb8W

2

5 Comments

Perhaps US citizens of Indian origin with needed skills may be able to join this new program? I presume, H1B folks may be out of purview of this opportunity I suppose! "..According to officials, the programme is being led by the Office of Personnel Management (OPM) along with other government agencies. Participants will be deployed across departments to help modernise systems, with a strong focus on areas such as artificial intelligence, software development, cybersecurity, data analytics and technical project management. What makes the programme interesting is the list of private-sector partners backing it. At least 28 technology companies are involved, including Apple, Amazon Web Services, Microsoft, Meta, Nvidia, OpenAI, Oracle, Uber and xAI, among others. These companies will not directly hire candidates at the start, but they will offer training support, mentorship and exposure through speaker sessions featuring top executives." https://lnkd.in/ggftfaJh

2