📍 **VA <> SF | 🎨 **Product Designer | ⚙️ Frontend Developer | 🤖 AI-native tools builder

I build products at the edge of design, infrastructure, and AI agents.

My work usually starts with the same question:

How do we help people do their best work with AI - without replacing their judgment, context, or craft?

Right now I’m building open-source and startup projects around agentic reliability, local developer workflows, AI-native infrastructure, and physical-world intelligence.

• 🛡️ Rivora - open-source adaptive reliability for teams using AI agents
• 🧭 Root - a safer, agent-friendly package workflow powered by Nix
• 📦 Fathom - understand any software package in seconds
• 🔐 Prax - every agent action explained before it happens

I’m building tools for the future where humans and AI agents share responsibility for production systems.

Rivora helps teams understand:

• what changed
• why something broke
• what risk exists now
• what action should happen next
• what evidence supports the recommendation

The goal is not to replace SREs, DevOps, or platform teams.

The goal is to give them better context, faster triage, safer automation, and receipts they can trust.

I’m interested in developer tools that make agents more useful, more inspectable, and more grounded.

That includes:

• local-first AI reviewers
• CLI-first workflows
• agent-readable docs
• deterministic receipts
• safer package installs
• local/cloud hybrid orchestration
• tools that help humans stay in the loop

Open-source adaptive reliability for modern infrastructure and AI-assisted engineering teams.

Built around:

• event timelines
• correlation
• root-cause receipts
• Slack-native workflows
• connector SDKs
• on-prem support
• human approval gates
• agent-readable reliability context

A safer package manager workflow for developers and coding agents.

Root is built around the idea that package installs should be:

• reproducible
• reversible
• inspectable
• agent-friendly
• powered by Nix without making users become Nix experts

Repo: github.com/sgr0691/Root

A developer tool for understanding software packages quickly.

The goal:

fathom hono

And in seconds, get a readable receipt explaining what the package is, how it behaves, where risk exists, and whether it fits your project.

Prax is about trust in agentic systems.

Every agent action should be explainable:

• what the agent wants to do
• why it wants to do it
• what files, systems, or data it may touch
• what risk exists
• what approval is needed

I believe AI products should make people feel more capable, not less involved.

The best tools do not remove the human.

They give the human:

• more context
• better timing
• clearer tradeoffs
• safer defaults
• faster paths from idea to reality

I like building products that feel practical, grounded, and useful before they feel magical.

I’m a design engineer with experience across:

• product design
• frontend engineering
• accessibility
• design systems
• AI-native workflows
• cloud infrastructure products
• developer tools
• early-stage startups

I’ve worked across design and engineering roles, including product work at companies like Code42, Elevate Security, Oracle, and SZNS Solutions.

Today, I’m focused on building tools that help people work with AI in ways that are more reliable, transparent, and human-centered.

• Design the system before polishing the UI
• Build small, testable loops
• Make the receipt visible
• Keep the human in control
• Prefer useful over impressive
• Ship, learn, tighten, repeat

• Building Rivora into a serious open-source reliability layer
• Making Root useful for local developer and agent workflows
• Exploring local AI reviewers and planners on Mac
• Writing clearer docs for AI coding agents
• Designing products where humans and agents collaborate well

You can connect with me on X or GitHub.

AI should not turn people into spectators. It should give them better leverage, better context, and better ways to act.