The first time I implemented Vamana from the DiskANN paper, my approximate nearest neighbor index was slower than brute force. On tiny test fixtures, brute force took 0.27 ms per query. My Vamana implementation took 22.98 ms. That sounds absurd. ANN exists to skip work. The problem was not the algorithm. It was how I mapped the paper's abstractions to actual data structures. The DiskANN pseudocode
Some time ago, I was building a chat application using AWS Websocket API gateway. Things were going smoothly. I created a WebSocket API Gateway, added $connect, $disconnect, and sendMessage/addGroup routes. From the frontend (React) side, everything was fire-and-forget. You send a message, and the onMessageHandler takes care of it 💪🏼 But then a new requirement of uploading files using S3 signed
When you have 5 unrelated questions, should you pack them into one message to the LLM, or send 5 requests simultaneously? Which is faster? Splitting into multiple independent parallel requests is almost always faster. This isn't a gut feeling — it's determined by the underlying inference mechanism of LLMs. Let's walk through the reasoning from first principles. To understand this problem, you firs