Software Engineer | AI Systems | Distributed Architectures

Welcome to my GitHub profile. I work on AI infrastructure, intelligent agents, distributed systems, and simulation platforms, with a strong focus on performance, scalability, and research-driven engineering. My projects span from LLM acceleration layers to evolutionary world models and peer-to-peer systems.

A production-grade, multi-tier caching system for Large Language Model embeddings and semantic search results, designed to deliver sub-millisecond latency for repeated queries.

Key Features

• Multi-layer cache hierarchy (in-memory, Redis, persistent vector DB)
• Intelligent cache promotion & eviction strategies
• Backend-agnostic design (Faiss, Qdrant support)
• Async-ready, type-safe Python APIs
• CLI utilities for ingestion, querying, and benchmarking

Tech Stack: Python, Redis, Vector Databases, AsyncIO

A scalable crawling and indexing engine built for RAG (Retrieval-Augmented Generation) pipelines. It converts web data into a locally searchable knowledge base and exposes clean APIs for integration with AI agents.

Highlights

• Distributed-friendly crawler architecture
• REST backend with Python SDK
• Designed for knowledge ingestion at scale
• Docker-ready deployment

Tech Stack: Python, FastAPI, Docker, Async Crawling

A research-oriented implementation of a visual world model that learns physical dynamics from video sequences and supports imagination-based reinforcement learning.

Focus Areas

• VQ-VAE for compact visual representation
• ConvGRU-based dynamics prediction
• Dreamer-style policy learning
• Apple Silicon (MPS) & CUDA support

Tech Stack: PyTorch, Reinforcement Learning, World Models

An advanced artificial life (ALife) simulation platform combining evolutionary algorithms, neuroevolution, and multi-agent reinforcement learning to study adaptive behavior in complex environments.

Project Highlights

• Hierarchical neural agents with metabolic constraints
• Genetic algorithms with inheritance & mutation
• Emergent tribal and survival dynamics
• Rich simulation logging & visualization

Tech Stack: Python, Evolutionary Algorithms, Multi-Agent Systems

A lossless peer-to-peer photo sharing web application using WebRTC data channels, enabling direct browser-to-browser transfers without cloud storage.

Key Features

• WebRTC-based P2P transfers
• QR-based room joining
• Real-time transfer statistics
• Mesh-ready signaling via Socket.IO

Tech Stack: TypeScript, WebRTC, Express.js, Socket.IO

• Languages: Python, TypeScript, Go
• AI / ML: PyTorch, Reinforcement Learning, Embeddings, World Models
• Systems: Distributed Caching, Web Crawling, P2P Networking
• Infrastructure: Docker, Redis, Vector Databases

Feel free to explore the repositories, open issues, or start a discussion if you’d like to collaborate or dive deeper into any of these projects.