▶ Videos →

📰 Dev.to · Mayank Ketkar

5 articles · Updated every 3 hours · View all reads

All Articles 131,250 Blog Posts 136,246 Tech Tutorials 33,996 Research Papers 25,453 News 18,664 ⚡ AI Lessons

One Week in Ray: 21 Bugs Between Us and a Production ML Pipeline

Dev.to · Mayank Ketkar 📊 Data Analytics & Business Intelligence 5mo ago

One Week in Ray: 21 Bugs Between Us and a Production ML Pipeline

The pipeline finished. All four checkpoints evaluated. Metrics written to cloud storage. Zero errors...

The Ghost in the Batch: How vLLM Silently Switches Algorithms

Dev.to · Mayank Ketkar 🧠 Large Language Models ⚡ AI Lesson 5mo ago

The Ghost in the Batch: How vLLM Silently Switches Algorithms

You run Qwen3-VL on a single prompt. You record the output logprobs to full precision. Then you run...

Two Ways to Move Tensors Without Stopping: Inside vLLM's Async GPU Transfer Patterns

Dev.to · Mayank Ketkar 🔢 Mathematical Foundations 5mo ago

Two Ways to Move Tensors Without Stopping: Inside vLLM's Async GPU Transfer Patterns

A single torch.cuda.synchronize() in the wrong place can erase every optimization you spent weeks...

Your Ray Data Pipeline Works at 10K Samples. Here's Why It Crashes at 1M.

Dev.to · Mayank Ketkar 🚀 Entrepreneurship & Startups 5mo ago

Your Ray Data Pipeline Works at 10K Samples. Here's Why It Crashes at 1M.

There's a moment every ML infrastructure engineer knows: the evaluation pipeline that worked...

Compiling the Vision Encoder: Squeezing 3% More Throughput from Qwen3-VL on Hopper GPUs

Dev.to · Mayank Ketkar 🧠 Large Language Models ⚡ AI Lesson 5mo ago

Compiling the Vision Encoder: Squeezing 3% More Throughput from Qwen3-VL on Hopper GPUs

When you run a vision-language model through vLLM, the framework does something clever: it compiles...