A three-node kubeadm cluster running on a mixed bare metal + VM environment, backed by a RAIDZ2 ZFS pool exposed over iSCSI through democratic-csi. MetalLB for load balancing, ingress-nginx with cert-manager and an internal CA, and a Prometheus/Grafana/Loki observability stack. Runs my own infrastructure day to day.

An OpenAI-compatible inference endpoint serving open-weight models from vLLM on a 16GB RTX 2000 Pro passed through to a VM. Covers quantisation and memory-budget tradeoffs, GPU passthrough under Proxmox, and exposing the service to my own applications over Tailscale.

In this project, we scrape data from Bayut.com to create a dataset of over 50 thousand listings to analyse real estate prices in Dubai. We visualized the average prices for each area using PowerBI and used this report to identify the overall market rates by neighborhood.

The aim of this project is to create a machine learning model that can classify URLs as ads or trackers.To this end, we have tried various machine learning algorithms to find the most promising algorithm for this project.

This project aims to visualise and analyse the effect of certain promotions on customer conversion. Using this information, the website can identify users with the most conversion potential and can understand the effects of different types of offers.

Analysis of dataset containing over 10,000 tweets to identify the sentiment expressed with the help of natural language processing, using Tensorflow and graphs to provide visual representation of collected data, achieving a sentiment classification accuracy of 92% using various Machine Learning algorithms such as neural networks, logistic regression, XGBoost and Random forests.

Analysis of dataset containing over 227,000 Amazon reviews to identify customer satisfaction with the product with the help of natural language processing, using Tensorflow and graphs to provide visual representation of the data, achieving sentiment classification of 90% and identified overall customer satisfaction with various products using various Machine Learning algorithms such as neural networks, regression, XGBoost , LightGBM, Random forests, SVM classifiers and hyper parameter tuning.