My Mission - Engineering the Unpredictable

From Safety-Critical to Context-Critical

For over a decade, I engineered safety-critical software for avionics and medical devices. In those industries, mostly working is a failure state. Systems must be robust, deterministic, and fail-safe.

Recently, I applied that rigor to a different domain: expert systems for decision-making. I built a prototype agent designed to act as a private coach for complex situations. Through that process, I discovered a fundamental limitation in the current AI stack.

Standard RAG is insufficient for expert-level reasoning. It retrieves text, but it fails to retrieve structure or hierarchy. It flattens wisdom into data.

This realization shifted my focus. I am no longer just building applications. I am re-architecting how we store, retrieve, and execute high-value context.

The Mission: Three Core Capabilities

I am developing three distinct capabilities at The Foundry. These are not just projects. They are infrastructure for the next generation of software architecture.

1. Judgment Packaging (The Intelligence Engine)

The Problem: Current AI models are excellent at syntax but struggle with deep context maintenance over long horizons. They hallucinate because they lack a map.

The Solution: I am moving beyond flat RAG architectures. I am building Mip-Mapped Knowledge Graphs - systems that store information at varying levels of resolution, from the one-sentence gist to the full source text.

The Goal: To demonstrate that you can extract domain expertise, structure it hierarchically, and deploy it as an agent that provides consistent, high-judgment guidance. This is the difference between a chatbot and a strategist.

Current work: https://github.com/andreirx/FRAKTAG

2. The Agency Operating System (AmodX)

The Problem: The web development industry is trapped in a Frankenstein Stack of legacy CMSs and plugins. This creates massive technical debt, security risks, and maintenance overhead that stifles creativity.

The Solution: A serverless, air-gapped operating system built on AWS Lambda and DynamoDB. It eliminates the idle tax of traditional servers and the security liabilities of PHP-based runtimes.

The Goal: To allow technical teams to deploy infrastructure that scales to zero cost when idle and handles enterprise traffic instantly. It treats content as structured data, making it natively visible to the new generation of AI search engines.

Current work: https://github.com/andreirx/AMODX

3. Private AI & Collaboration Infrastructure

The Problem: Some organizations need the power of Generative AI but cannot tolerate data leaks. They need on premises AI systems.

The Solution: private AI deployments, building organization specific applications, chaining LLM "lego blocks", making them testable against various local-running models and against various tasks.

The Goal: Bring generative AI "in house" for such enterprises. The architecture should handle the demands of regulated enterprise workflows.

FRAKTAG is also playing a part here.

The Philosophy

My work is driven by curiosity, solving difficult problems, and the idea of connecting high-fidelity minds with high-fidelity tools.

And I do have an eye on education too, as in making real educational tools for school children to help them learn different topics at different levels.