I’m a data engineer building context infrastructure for reliable AI systems.
My work focuses on the systems around the model: semantic data layers, scoped context, agent workflows, evals, guardrails, and tooling that helps AI systems use the right information at the right time.
AI does not need more magic. It needs better systems.
Contact • X • LinkedIn • GitHub • Substack • SoyPete Tech YouTube • Domesticating AI • Twitch
I am working on tooling and experiments for context engineering.
To me, context engineering is not prompt management. It is the infrastructure that decides what information an AI system can access, how that information is scoped, how business meaning is represented, and how agent behavior is tested before we trust it.
Current areas of focus:
- semantic data layers
- ontologies and knowledge graphs
- scoped context and access boundaries
- agent workflows and tool use
- evals and reliability testing
- self-hosted AI infrastructure
- data systems for AI
The long-term goal is to build practical tooling for teams that need AI systems to work with real business context without turning into unreliable automation.
Go libraries and experiments for working with ontologies, semantic models, and graph-backed validation.
This project supports my larger work around context engineering: representing business meaning outside the model so AI systems can reason over constrained, inspectable structures.
Agent middleware experiments for tool use, workflow control, context management, and reliability testing.
Pedro started as a self-hosted AI bot, but the project has grown into a place where I test ideas about agent behavior, guardrails, and system design.
Experiments with graph-backed semantics and evals for validating AI behavior.
The goal is to treat agent behavior more like software behavior: observable, testable, and constrained.
I share the work through SoyPete Tech and Domesticating AI.
- SoyPete Tech Substack — writing on context engineering, AI reliability, data systems, and practical AI engineering
- SoyPete Tech YouTube — technical videos, build-in-public streams, and practical AI engineering content
- Domesticating AI — podcast on practical AI for developers
- Twitch — live software development and building in public
- Discord — community discussion
- Contact hub — speaking, collaboration, sponsorship, podcast appearances, and tooling conversations
This is not AI thought leadership. This is building the systems around the model.
I have worked as a data and backend engineer building data systems, APIs, services, and cloud-native infrastructure.
My background includes:
- data engineering and reliability
- backend software systems
- semantic data modeling
- AI systems and self-hosted models
- developer tooling
- Go services and education
- community organizing in Utah’s software ecosystem
I have worked at companies including Nav, Weave, Tailscale, and SchoolAI.
I have also created Go courses and workshops, including:
These projects are part of my teaching archive. Go is still one of the tools I reach for when I need boring, reliable software. It is no longer the center of my public work, but it is still part of how I build.
These repos are part of my teaching archive.
An introductory Go course for people with less than one year of programming experience.
Exercises for a Production Go workshop focused on patterns, anti-patterns, and memory management.
Exercises for a Go web services workshop.
Example project from a Go web development course.
For speaking, collaboration, sponsorship, podcast appearances, or context engineering tooling conversations:
- Contact hub: SoyPete Tech Linktree
- LinkedIn: Miriah Peterson
- Substack: SoyPete Tech
- GitHub: Soypete
- X: @captainnobody1