FPGA digit convolution accelerator
A hardware convolution accelerator that detects the digits 8 and 9 in a 64×64 grayscale image, driven over AXI4-Lite from a Zynq PS and deployed to a remote FPGA board.
B.Sc. Computer Engineering · Shahid Beheshti University
I build things that live where software meets silicon — embedded systems, IoT and edge computing. My work runs from Verilog on FPGA fabric and model-based real-time controllers up to on-device inference and the networks that carry it.
Four semesters as a teaching assistant across advanced programming, operating systems and computer architecture. Currently in Tehran, Iran.
The problems I keep coming back to.
Firmware, real-time control and model-based design on ARM Cortex-M, STM32 and ESP32.
Sensor networks, MQTT telemetry and dashboards that make a fleet of devices legible.
Pushing the decision to the device — latency, privacy and bandwidth budgets that hold.
Small models on small silicon: quantised inference where there is no cloud to call.
Simulating a plant well enough that the model, not the hardware, catches the bug first.
Protocol behaviour under load, and the gap between the RFC and the wire.
Cryptographic primitives in hardware, key management, and building things that fail closed.
I would like to hear about it.
Get in touchB.Sc. in Computer Engineering · Tehran, Iran
GPA 18.43 / 20.
Diploma, Mathematics & Physics · Tehran, Iran
GPA 19.39 / 20.
Ranked in the top decile of a computer engineering cohort of roughly 100 students.
Rank 71 among more than 37,000 participants in the regional quota.
Selected work — all of it open on GitHub.
A hardware convolution accelerator that detects the digits 8 and 9 in a 64×64 grayscale image, driven over AXI4-Lite from a Zynq PS and deployed to a remote FPGA board.
An AIoT pipeline that classifies motion from simulated MPU6050 sensor data: inference runs at the edge through Edge Impulse WebAssembly, results travel over MQTT, and ThingsBoard handles live monitoring and alerts.
A smart-home system joining KNX devices to ThingsBoard, with MCP and a locally hosted Ollama model giving it Persian natural-language control.
A multi-room heating controller modelled end-to-end under the V-model: thermostat logic, a safety checker and room thermal dynamics in Simulink and Stateflow, validated with Simulink Test and Embedded Coder.
Two-key 3DES in Verilog, built three ways — single-cycle, multi-cycle and pipelined — with ECB, CBC and OFB modes, so the area/throughput trade-off is measurable rather than argued.
Modular exponentiation in hardware, assembled from Montgomery multiplication, Karatsuba multiplication, non-restoring division and the extended Euclidean algorithm.
The Tiny Encryption Algorithm implemented on remote Mitra Z10/Z20 boards, wired to the processing system through AXI4-Lite and built with Vivado and the Xilinx SDK.
A MIPS processor in Verilog, down to the ALU and memory subsystem — the whole datapath rather than a block diagram of one.
An educational OS simulation with a CPU, MMU, threads and a file layer, used to study page-replacement, virtual memory, synchronisation and scheduling under measurable fault behaviour.
A two-phase cryptography project: symmetric encryption, RSA and DSA, a small PKI with certificate management, and a client–server channel built on top of it.
A minimax agent whose heuristic weights were tuned by a genetic algorithm. Third place in the course AI competition.
A logic function built from transistor primitives rather than gates — the level where the abstraction stops being free.
Nothing in this category yet.
Teaching assistance, academic mentoring and service.
Dr. Seyed-Hosein Attarzadeh-Niaki · Shahid Beheshti University
Dr. Mojtaba Vahidi-Asl · Shahid Beheshti University
Prof. Mohsen Ebrahimi Moghadam · Shahid Beheshti University
Dr. Dara Rahmati · Shahid Beheshti University
Dr. Monireh Abdoos · Shahid Beheshti University
Dr. Yasser Shekofteh · Shahid Beheshti University
Dr. Mehran Alidoost Nia · Shahid Beheshti University
Dr. Mahmood Neshati · Shahid Beheshti University
Dr. Haghighi · Shahid Beheshti University
Supervised by Prof. Shams Aliee
Supervised by Dr. Salimi Badr
Open to research collaboration, graduate opportunities and interesting hardware.