University of Tehran · HPNL

Researcher · Engineer · Builder

Ali Ghorbani Bargani.

I work where learning meets systems — from large language models and representation learning to FPGA security, computer architecture, and real-world AI products.

Ali Ghorbani Bargani
Computer Engineering Learning × Hardware × Systems
Current focus LLMs · Low-label learning
Also FPGA security · AI systems
Scroll
Machine LearningLarge Language ModelsRepresentation LearningFPGA SecurityComputer SystemsApplied AI
01About

A research mind with an engineering bias.

I am an M.Sc. student in Computer Engineering at the University of Tehran, specializing in Computer Architecture and working with the High Performance Network Laboratory (HPNL).

I care about clean scientific questions and systems that work outside a notebook. My recent work spans low-label learning, LLM-derived representations, FPGA hardware security, network measurement, and product-oriented AI.

I completed my B.Sc. in Computer Engineering at Shahid Bahonar University of Kerman (2020–2024).

Based inTehran, Iran
LabHPNL
Ali Ghorbani Bargani in an academic meeting
Research / Discussion / Collaboration
02Selected work

Work that crosses the boundary between research and building.

Selected projects. The complete archive lives on GitHub.

01Real-world productFeatured

BeshmarAI

Private, on-device AI for pill counting.

A browser-based AI application designed for practical pharmacy workflows. Inference runs on the user's device with adaptive GPU/WebGPU or CPU/WebAssembly execution, keeping the workflow fast, inspectable, and privacy-aware.

On-device AIComputer VisionWebGPUWebAssemblyPWA
beshmarai.ir
BeshmarAI pill counting interface
BeshmarAIOn-device AI.
Built for real use.
Neural Network Processing Unit GitHub project preview
02Hardware / AI

Neural Network Processing Unit

Verilog processing element for MLP operations and neural-network hardware acceleration.

Repository
Adversarial attacks and Vision Transformers project preview
03Machine Learning

Adversarial Attacks & Vision Transformers

Robustness experiments with ViTs, ResNet, FGSM, adversarial training, and Grad-CAM.

Repository
Fast-SCNN project preview
04Computer Vision

Urban Scene Segmentation

Fast-SCNN on CamVid for efficient semantic segmentation and real-time scene understanding.

Repository
DNS protocol performance project preview
05Network Systems

DNS under Packet Loss

Controlled comparison of Plain DNS, DoT, and DoQ under 0–20% packet loss.

Repository
03Research

Questions first. Machinery second.

I am interested in representation quality, label efficiency, intelligent systems, and the hardware beneath them.

01

LLMs & Representation Learning

Low-label learning, compact response representations, and empirical questions about what useful structure language models expose.

02

FPGA & Hardware Security

Remote power side channels, shared FPGA environments, active-fence countermeasures, and secure reconfigurable computing.

03

Systems & Measurement

Computer architecture, hardware acceleration, network measurement, benchmarking, and performance evaluation.

2026

Publication · IEEE DCHPC

Evaluating the Impact of Active Wire Fence Placement on Remote Power Side-Channel Attacks in Multi-Tenant FPGAs

Sina Ojaghi · Ali Ghorbani Bargani · Sana Babayan Vanestan · Mona Hashemi · Siamak Mohammadi · Ahmad Khonsari

DOI / Publication
Ali Ghorbani Bargani at a snowy mountain viewpoint
Outside the lab Distance changes perspective.
04Journey

Education & experience.

Teaching

Teaching Assistant

Fundamentals of Distributed Systems · University of Tehran

Ali Ghorbani Bargani
Build.Test.Learn.
Tools I use
PythonPyTorchscikit-learnOpenCVVerilogVHDLFPGAVivadoModelSimLinuxGitJavaScriptWebGPUWebAssembly
05Contact

Have a research question or a good problem?

Let’s talk.

Research collaboration, engineering, AI systems, or product work — email is the simplest way to reach me.