Junaid Rehman

Junaid co-founded TwoApps in 2025 and works remotely from Islamabad. His focus is hardware-aware AI: spiking neural networks that map cortical-column dynamics to silicon-ready models, and memory layers aimed at very low-power edge inference. He also brings production ML engineering — document pipelines, LLM fine-tuning and on-device deployment — and a background in analog/mixed-signal chip design.

• Spiking neural networks: LIF neurons, surrogate gradients, STDP, event-based vision
• snntorch, Norse, Intel Lava (Loihi) and sPyNNaker (SpiNNaker2)
• Analog/mixed-signal IC design: Cadence Virtuoso, Spectre, GDSII layout, PEX
• PLLs, op-amps and switched-capacitor circuits on TSMC 65nm and SKY130
• Verilog/SystemVerilog, RTL, synthesis, STA and FPGA
• Production ML: OCR and document pipelines, LLM fine-tuning, on-device deployment
• Python, C++, PyTorch, Docker, Linux

• B.S. Electrical Engineering (Electronics) — PIEAS, Islamabad, 2017–2021. Thesis: fabrication and characterization of FET and diode devices using transition metal dichalcogenides (TMDs).
• Reviewer, NeurIPS 2026 — Academic service