Arbaz Khilji

I’m a backend-focused software engineer who builds secure, real-time, concurrent chat systems from scratch using Java. I design custom application-layer protocols and scalable multithreaded server architectures without frameworks.

In my Secure Multithreaded Chat Server, I engineered TLS-secured client-server messaging using X.509 PKI and a thread-pool–driven concurrency model with backpressure via connection limits. I added lock-free concurrent registries for session and user management and built explicit observability hooks for monitoring connections, commands, and system health.

I intentionally design for evolution: my chat backend is chatbot-ready, so AI agents and LLM-based responders can plug in without destabilizing the core transport layer. I’m currently studying B.Tech in Artificial Intelligence and Data Science, which supports my shift toward conversational systems.

Now, in my LLM-Powered Chatbot Service (Java–Python Integration), I integrated a Python-based LLM chatbot with my custom Java real-time chat server. I built a language-agnostic protocol bridge, added session-aware conversation memory, orchestrated prompts for safe responses, and kept the AI service behind backend-only boundaries for security—while targeting low-latency responses and future token streaming.