Md. Mahmudul Kabir Himel
I enjoy turning curiosity into things that work. I research and explore AI, cybersecurity, and emerging technologies, always learning, tinkering, and occasionally breaking things along the way.
Firewall traffic monitor, live
A simulated packet stream hitting a firewall โ each packet is scored against a set of traffic-shape rules (unusual size, odd port, bursty timing), and flagged as normal or malicious in real time. It's a toy version of the kind of anomaly-detection logic behind intrusion detection systems, the same space my thesis work sits in.
Neural network forward-pass, live
A small feedforward network โ input, two hidden layers, and an output layer classifying traffic as normal, suspicious, or malicious. Each cycle, a random input vector is pushed through: watch the signal pulses travel edge by edge, node glow track activation strength, and edge thickness/color track weight magnitude (teal/green = positive, red = negative). A simplified picture of the kind of network behind intrusion detection models.