⠶⠶⠶⠶⠶⣶⣶⣶⣶⣶⣤⣤⣤⣤⣤⡤⣀⣀⣀⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⣀⣀⣀⡠⢤⣤⣤⣤⣤⣤⣤⣔⣒⣶⣶⣶⣶⣶ ⣀⣀⣀⣀⣀⣀⣀⣠⣤⣤⣤⣤⣤⣬⣭⣭⣭⣉⣙⣓⣛⣛⣻⣶⣶⢶⣤⣤⣄⣀⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⣀⣠⣤⣤⣲⣶⣶⣽⣛⣛⣓⣾⣯⣭⣭⣭⣭⣭⣤⣤⣤⣤⣤⣤⣤⣤⣄⣀⣀⣀ ⠛⠋⠉⠛⠛⠛⠛⠛⠛⠿⠿⠿⠿⠿⠿⣿⣭⣭⣭⣭⣭⣭⢍⣉⣒⣛⣛⣛⣛⠛⠿⠿⠿⠷⣶⣦⣤⣄⣀⠀⠀⠀⠀⠀⠀⣀⣠⣤⠴⠶⠶⠛⠛⠛⠋⢉⣉⣉⣉⣀⣀⣀⣤⣤⣤⣤⣤⣤⣤⣶⡶⠶⠶⠶⠶⠶⠾⠟⠛⠛⠛⠛⠛⠛ ⠶⠶⠶⠶⠶⢖⣒⣒⣒⣒⣒⣒⣒⣒⣒⣒⣒⣒⣒⣒⣒⡻⠿⠿⠿⠿⣿⣿⣿⣿⣿⣿⣿⣿⣿⡿⠶⢾⣿⣿⠿⠖⠲⠿⣿⣷⣶⡶⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⠿⠿⣛⣛⣛⣓⣒⣒⣤⣤⣤⣤⣤⣤⣤⣤⣤⣤⣤⣤⣤⣤⣤⣤⣤⣤ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⡠⣤⣴⢶⣮⣻⣿⡻⠿⠟⠛⢋⣉⣭⣤⠶⠞⠋⠉⠁⠀⠀⠀⠀⠀⠀⠈⠉⠙⠻⠷⣶⣮⣭⣝⣛⡫⠿⢟⣿⢟⣷⡶⣦⣤⣄⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⣤⣴⠶⠝⣓⢯⣝⡶⠽⠓⠋⣉⣠⣤⣶⡾⠟⠛⠉⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠉⠛⠻⢿⣿⣶⣶⣭⣍⣚⠫⢖⣫⠽⣛⠻⠶⣦⣤⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⣀⣠⣔⡶⠟⠛⣉⢥⣔⡮⠟⢚⣉⡤⢴⣲⣯⡻⠟⠓⠉⠀⢒⠒⠤⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣠⡴⠶⠒⠲⡍⠚⠳⢟⣽⣛⡷⢶⣭⣕⣫⠵⣢⠬⣉⠛⠳⢶⣤⣀⡀⠀⠀⠀⠀⠀⠀ ⠀⣀⣠⣴⠶⠟⠋⢉⡠⣔⡶⢽⣚⣭⠴⢖⣫⣭⡲⠟⠓⠉⠀⠀⠀⠀⠀⣿⣿⣿⣿⣶⣭⣑⠢⣄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⣤⢖⣫⣽⣶⣿⣿⣷⠀⡇⠀⠀⠀⠀⠉⠚⠳⢶⡭⣙⡛⠷⢿⣵⣯⣖⡢⢄⡉⠙⠳⠶⣤⣀⠀⠀ ⠿⢯⣭⣀⢤⣲⣾⣷⡯⠗⠚⣉⢥⡶⠮⠓⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⡄⢻⣿⣯⠀⠈⠙⠻⢿⣶⣭⣒⠤⣀⠀⠀⠀⢀⣤⠶⣻⣵⣾⠟⠋⠉⠀⢹⣿⡟⣰⠇⠀⠀⠀⠀⠀⠀⠀⠀⠈⠉⠛⠷⣢⢤⣉⠙⠻⠿⣾⣵⣒⡤⣀⣉⣽⡷ ⣤⣶⣽⠿⠻⠿⣭⣤⣴⠾⠛⠊⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⢆⠻⣿⣧⡀⠀⠀⠀⠈⠻⣿⣷⣦⣹⣶⣟⢋⣴⣿⡿⠋⠀⠀⠀⠀⣠⣿⠟⣰⠋⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠉⠙⠻⢖⣤⣄⣨⡽⠿⠿⢿⣶⣤ ⠉⢛⣶⣤⣶⣟⠋⠉⠛⠷⢶⣤⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠣⡙⢿⣿⣦⣄⡀⠀⠀⢘⣿⣿⢿⣅⣸⣿⣿⡏⠀⠀⠀⢀⣴⣾⡿⢣⡺⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⡠⣤⠶⠛⠋⣛⣷⣶⣶⣿⠟⠋ ⢞⣛⠉⠀⠉⠚⠽⣳⢶⡤⣄⡈⠙⠻⠶⣦⣄⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠘⠢⡙⠻⣿⣿⢿⣿⣿⢟⣿⠈⡿⡟⡽⣿⣿⣷⣾⣿⡿⠟⣩⡾⠋⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⢤⡲⠞⢓⣉⣤⣴⡾⢿⡻⠝⠋⠉⠙⣻⡷ ⠀⠀⠉⠒⠢⠤⣀⠀⠉⠚⠳⢞⡭⣒⠢⠤⣉⡛⠻⢶⣤⣀⡀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠑⢄⡉⠙⠛⠛⠛⠁⢠⠇⡷⣝⠮⠭⠉⣉⣥⡴⠟⠃⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⢤⡴⠮⣛⣩⣵⡶⠿⠛⣋⠭⠒⠉⠁⣀⡤⠔⠚⠉⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠉⠑⠒⠤⢄⣈⠑⠛⠵⣢⡬⣉⠒⠪⠭⣛⣷⢦⣤⣀⠀⠀⠀⠀⠀⠀⠀⠈⠙⠒⠒⠒⠒⠋⠀⠙⠚⠉⠙⠛⠛⠉⠀⠀⠀⠀⠀⠀⠀⠀⣀⢤⡴⣮⣻⣭⡷⠾⠛⠋⣉⠤⠔⠊⣉⣠⠤⠖⠚⠉⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠉⠑⠒⠤⢌⣑⡛⠵⢖⣤⢌⣉⠛⠛⠿⢷⣶⣤⣀⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⣤⣴⣾⣿⠿⠿⠛⠋⣉⡠⠤⢒⣊⣩⡤⠶⠚⠋⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠉⠑⠒⠫⠵⠿⣳⣶⣦⣤⣌⣉⣛⠛⠶⢦⣤⣀⡀⠀⠀⠀⠀⠀⠀⢀⣀⣠⣴⡶⠿⠟⠛⢋⣉⣁⣤⣔⣶⣮⠽⠶⠚⠛⠉⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⠛⣛⣛⣛⣛⣒⣒⣚⣻⣿⣿⣿⣿⣿⣷⠶⢾⣿⣿⣶⠦⠴⣶⣿⡽⠿⢷⣶⣶⣾⣿⣿⣿⣟⣛⣛⣛⣛⣛⣛⣛⠋⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⠙⠛⠛⠛⠛⠛ ⣤⣄⣀⣀⣀⣤⣤⣤⣤⣤⣴⣶⣶⣶⣶⣒⣛⣛⣛⣛⣛⣛⣉⣉⠭⣭⣭⣭⣥⣤⣶⣶⣶⠶⠿⠟⠛⠋⠉⠀⠀⠀⠀⠀⠀⠉⠙⠛⠲⠶⢶⣤⣤⣤⣌⣉⣉⣉⣉⣉⠉⠛⠛⠛⠛⠛⠻⠿⠶⠶⠶⠶⢶⣦⣤⣤⣤⣤⣤⣀⣀⣀⣀⣀ ⠀⠈⠉⠉⠉⠉⠉⠉⠉⠉⠉⠉⣉⣉⣉⣉⣉⣩⣭⣭⣭⣭⣿⠿⠿⠾⠛⠛⠋⠉⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠉⠉⠛⠛⠻⠷⠿⢿⣯⣭⣭⣭⣍⣉⣛⣛⣛⡛⠛⠛⠛⠛⠛⠋⠉⠉⠉⠉⠉⠉⠉ ⠿⠿⠿⠿⠿⠿⠟⠛⠛⠛⠛⠛⠛⠛⠋⠉⠉⠉⠉⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠉⠉⠉⠙⠛⠛⠛⠛⠻⠿⠿⠿⠿⠿⠿⠶⠶⠶⠶⠶
Double major in EE and physics, KFUPM. MSc ECE, KAUST.
Multi-Stream Partial Response Signaling. Two bits per symbol, interference added on purpose, and a turbo loop that untangles it. The one-dimensional convolutional sub-family of the FTN-inspired Nyquist Signaling Modulations; the coded receiver paper is not yet published. coded-msprs · NSM paper, arXiv:2511.08553 · MSc thesis
Integer inference, end to end. One network trained in PyTorch, quantized by hand, and implemented three times: Python, C++ on a Cortex-M7, and SystemVerilog on an FPGA. All three produce the same bytes on the same input, checked against committed golden vectors. nano-inference