AI and cybersecurity, mostly at the join between them. Finishing an MSc in Artificial Intelligence at Brunel University London (September 2026), after an MSc in Cybersecurity and a stint doing vulnerability assessment and incident response.
The work I find most interesting is the question of whether a model can be trusted, not just whether it scores well: what it does on inputs it has never seen, whether it gives the same answer twice, and whether anyone could reconstruct why it decided what it decided.
seoul-bike-demand-forecasting
Random Forest against a feed-forward network on a year of hourly rentals. The forest won, but the finding that mattered was a missing tf.random.set_seed that left R² swinging between −167 and −0.35 across identical runs, and a snowfall reading 3.7 cm outside the training range that pushed the network's output to 262,434 bikes. Includes a script that recomputes every reported figure from the saved predictions.
Python · scikit-learn · TensorFlow · pandas
Working through the OWASP Top 10 for LLM Applications and MITRE ATLAS, and running Garak against local models to build up the AI red-teaming side properly. Repos to follow.
- Email: hamzakazih1@gmail.com
- LinkedIn: https://www.linkedin.com/in/hamza-kazi-204175190/
