Throughout my career I’ve continuously explored emerging tools that enhance system capabilities and security postures. Beyond my enterprise work, I maintain a hands-on approach to new technologies by regularly experimenting with innovative open-source projects. Below are some recent tools I’ve implemented, modified, or studied—each offering unique solutions to complex problems in the technology landscape. Its also interesting to see how projects move though their lifecycles and how user input help shape their trajectory.
LocalAI is a self-hosted, lightweight alternative to OpenAI that runs on local hardware (in theory also on cloud instances). It offers a drop-in replacement REST API for AI models like GPT, LLaMA, LLaMA v2, and others, focusing on privacy and local execution.
Experience:
I implemented LocalAI as an on-premises solution for sensitive data processing where cloud-based AI services weren’t appropriate. By fine-tuning models for specialized analysis tasks, I created a secure environment for analyzing proprietary code and documentation without external data exposure. This implementation provided AI capabilities while maintaining complete data control and privacy. Good use for my local Nvidia 4090
vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. It significantly accelerates inference through PagedAttention, an algorithm that effectively manages attention key and value tensors in non-contiguous memory space.
Experience:
I experimented with vLLM to optimize throughput and latency when serving both smaller and larger language models. I achieved about double throughput compared to the same models on other local llm systems. I explored scaling models across multiple servers allowing for more parallel operations and allow optimization of multiple models layout in memory.
FreePBX is an open-source GUI that controls and manages Asterisk, the world’s most popular open-source communication server. It provides a web interface to simplify the process of building and managing complex telecommunication systems.
Experience:
Since 2007, I’ve leveraged FreePBX to architect and implement a comprehensive phone/web reservation system that was successfully deployed across 20 locations. By integrating custom modules and extending the platform’s capabilities, I created a unified communications solution that streamlined customer interactions and internal processes. The system incorporated IVR (Interactive Voice Response), custom call routing, and web-based management interfaces, demonstrating enterprise-grade telecommunications infrastructure using open-source technology. 99.999 uptime on both on prem and cloud based instances
LocalStack is a cloud service emulator that runs in a single container on your local machine or CI environment. It provides an easy-to-use test/mocking framework for developing cloud applications without connecting to cloud providers.
Experience:
Since 2017, I’ve integrated LocalStack into my AWS development workflows to streamline testing and quality assurance processes. By implementing local emulation of EC2, S3, Lambda, and other AWS services, I significantly reduced development cycles and testing costs while maintaining high confidence in cloud deployments. I’ve created comprehensive test suites and CI/CD pipelines that leverage LocalStack to validate complex cloud architectures before production deployment, ensuring seamless operation while eliminating costly cloud resources during development phases.
Electron is a framework for creating native applications with web technologies like JavaScript, HTML, and CSS. It enables building cross-platform desktop applications using a single codebase that works on Windows, macOS, and Linux.
Experience:
From 2012-2017, I leveraged Electron to develop a sophisticated host-based anomaly detection system for user behavior profiling. This cross-platform solution monitored system events, user interactions, and application usage patterns to establish behavioral baselines and identify potential security threats. By combining Electron’s native system access capabilities with advanced machine learning algorithms, I created a lightweight yet powerful security solution that could detect insider threats and unauthorized access with minimal system overhead.