Brian Oktavec
Projects

Tools and systems I’ve built or am working on.

From an internal platform and offline inspections to local AI tools, secrets management, and early-stage infrastructure and agent-security projects. Each page says what’s working and what’s still in progress.

Internal Operations Platform

PLATFORMWORK
manual workflows moved into one automated platform

As lead developer, I helped replace a manual, spread-out process with one internal application and automated many of its workflows. The team went from the 50s to the mid-20s while maintaining the same efficiency.

Internal ToolsWorkflow AutomationProduction Ownership
Case study →

Atlas

AI / LLMPERSONAL
private documents, searchable with local AI

A working local RAG assistant for documents like bills, medical records, and receipts. It retrieves relevant records to answer questions, says when the documents don’t have an answer, and uses SearXNG for unrelated questions.

RAGQdrantLocal ModelsSearXNG
Case study →

Inspectra

PLATFORMWORK
inspection work that keeps going offline

I led development across the product, including its identity service and an offline workflow that lets inspectors complete inspections without a reliable connection, then sync their work when they’re back online.

Offline PWAIdentityAutomatic Sync
Case study →

Keyveil

SECURITYPERSONAL
local-first secret access for coding agents

A macOS app, CLI, and MCP Bridge for giving local tools access to secrets without putting the secret value directly in the agent’s prompt. The Raycast extension is still in progress.

SwiftKeychainCLIMCP
Case study →

MailForge

AI / LLMPERSONAL
local AI email sorting and cleanup

A working email workflow that uses local AI to categorize messages and move them into folders. For example, GitHub CI mail goes to Development, where read messages older than 30 days are deleted.

Local AIEmailAutomation
Case study →

StackSmith

PLATFORMPERSONAL
early-stage Go tooling for safer Terraform workflows

An early-stage project for turning a small infrastructure specification and platform standards into reviewed Terraform. The S3 module foundation is in place; strict parsing, validation, and generation are still being built.

GoTerraformAWSInfrastructure as Code
Case study →

Sentinel

SECURITYPERSONAL
design-stage security controls for coding agents

A design-stage project exploring how to apply existing security policies to coding-agent tool calls. The goal is to add enforceable guardrails around what an agent can do, without duplicating the scanners and checks teams already use.

AI SecurityPolicyDeveloper Tools
Case study →