API cookieless AI Rust ad server, how?

[JosiahMaybe]

I know I can use Burn AI in Rust, possibly MiniLM, a geolocation database free, a geolocation API free for 1M calls a month, just curious how you put in like context, description of ads, and come out with good targeting? I kind of want Rust because safety, reliability, and speed. I know ONNX can be imported to Burn. How do models generally work and is it possible to get out like an array of numbers for rankings of fit for ads? Do I need to make that myself? This would be for games and things with no native ads SDK. Also, should all the things have a text description or what should I do because video, audio, image, and text are formats to try?

Just general advice how to use AI or do this? The geolocation database is a bit big. Any recommendations of small fast models that can understand stuff in many languages? Basically what I want is all data into the language model, output ranks of top few matches like an array, should be like a few milliseconds response time. Also, is this recommended to stay separate from a game server? This would be a cookieless ad server with an API, mostly for games, but other things are possible. X E.

 

[Zoeydaniels]

Building a cookieless AI ad server in Rust is a smart choice for maximizing speed and handling massive traffic without a garbage collector. While I usually focus on client-side optimization like tweaking an rbx fps unlocker to remove local frame bottlenecks, server-side performance requires a very similar mindset. Relying on context-aware data instead of tracking cookies should keep your pipeline incredibly fast. Good luck with the build!