Interview
summary
A Scala library that scores ONNX models on the JVM and integrates with existing Java/Scala services, built and deployed at Stripe, lets teams train in Python but run inference without adding Python services. The project parses ONNX graphs, maps operators to JVM implementations, and accepts real traffic in production, reducing deployment complexity and latency compared with shipping Python containers. Work aligns with interest in Scala 3 optimizations and avoiding Python services
AI-written summary · about
related talks
32mMachine Learning Infrastructure at Stripe: Bridging from Python JVM
45mLeverage Power of Machine Learning with ONNX
05mONNX Runtime
04mInterview
30mScalable ML acceleration with ONNX Runtime
41mLeverage Power of Machine Learning with ONNX
06mInterview
45mMaking Machine Learning Applications Fast and Simple with...