An intelligent ad blocker that learns from your browsing using LLM classification + hierarchical trie caching for instant lookups. This real-time ad blocking proxy intercepts HTTP/HTTPS traffic, classifies URLs using Groq AI (Llama 3.1 8B) on first visit, and caches decisions in a trie for microsecond future lookups. Unlike traditional blocklists, it creates a personalized, self-learning ad blocker that gets smarter over time.
A decentralized poker game engine written in Golang and Solidity. Built for trustless peer-to-peer gameplay, it implements TCP handshakes for reliable network communication and cryptographic state confirmation to ensure fair play without a central authority. The engine handles game logic, player actions, and consensus mechanisms for a fully distributed poker experience.
A machine learning system for predicting tennis match outcomes with custom-built Random Forest and Decision Tree algorithms implemented from scratch. Features advanced feature engineering including ELO ratings, player statistics, and match conditions. The model integrates with sportsbook odds APIs to identify value plays where predicted probabilities diverge from market odds, enabling data-driven betting decisions. Optimized with tuned hyperparameters (max_depth=15, min_samples_split=20), feature importance-based sampling, and weighted ensemble voting for prediction accuracy.