Projects

Design, launch, and monitor intelligent agents with precision. A full-stack platform for building and orchestrating AI workflows with visual flow editors, real-time monitoring, and seamless LLM integration.

Smart inference routing for Kubernetes clusters. Automatically routes LLM requests to optimal backends based on model availability, load, and latency. Supports vLLM, TGI, and other inference servers.

PDF annotation and streaming tool designed for content creators. Annotate documents privately, then stream your annotations live with a clean presenter view. Perfect for tutorials and live coding.

Complete infrastructure-as-code for managing my K3s and Harvester clusters. Includes ArgoCD deployments, Flux configurations, monitoring stack, and automated CI/CD pipelines.

Collection of optimized Docker images for running AI/ML workloads including vLLM, text-generation-inference, and other inference servers. Configured for both NVIDIA and AMD GPUs.

Kubernetes manifests and Helm charts for deploying vLLM inference servers at scale. Includes autoscaling, monitoring, and multi-model serving configurations.

Python framework for building AI agents with atomic, reusable components. Features tool calling, structured outputs, and integration with multiple LLM providers including local models.

Automated news collection and analysis pipeline using LLMs. Aggregates from multiple sources, extracts key insights, and generates summaries with sentiment analysis.

Shell scripts and utilities for managing Rancher Fleet deployments across multiple K3s clusters. Includes backup tools, health checks, and deployment automation.

My dotfiles and development environment setup. Includes Neovim config, Zsh customizations, tmux settings, and automated installation scripts.

A side project exploring game development with the Godot engine. Learning GDScript and experimenting with 2D game mechanics and procedural generation.