Do you think Python is too slow for high-throughput systems? Well, think again! In this talk, we’ll challenge the well-accepted belief that performant systems must be written in C++ by showcasing a Python-based architecture that handles thousands of messages per second. We’ll take a C-style approach to Python performance: aggressive memory budgeting, minimized re-allocations, and a hot path where every line of code matters.

We’ll explore the design and performance journey of a high-volume sensor network simulator that ingests and processes thousands of UDP messages per second from virtual devices scattered across London. These sensors emit temperature, humidity, gas, flood, and smoke data, which is often malformed or come in unpredictable bursts.

By the end of the talk, you’ll see how a carefully engineered Python stack can meet the latency and throughput demands typically reserved for C++. It’s not just “fast for Python” – it’s fast, readable, and it might just change your mind!