Soufiane Tahiri

Designed and developed tinyinfer, a neural network inference engine in pure C99 targeting resource-constrained embedded systems (ESP32: 240 MHz, 512 KB SRAM, 4 MB flash)

Implemented Dense, ReLU, and Softmax layers with float32 and int8 quantization support; zero dynamic memory allocation at runtime

Built a host-first validation pipeline to ensure numerical correctness before deployment on target hardware

Published as an open-source project: github.com/Soufiane-Tahiri/tinyinfer