piniq.ai

// Independent AI & Security Engineering

Private AI infrastructure and red team engineering.

Agents on your machines. Local models by default, your own API keys when you want them. Audits, red team engagements, and the environment they run in.

01

CHA0S AOE

Agentic operations environment: a native Rust desktop where agents run your shell, files, browser and databases alongside you. The environment our engagements run in.

macOS binary · $149/year · 177-crate Rust workspace

02

phylactery-mcp

A Rust MCP server that turns a Postgres database into memory: 768d vectors, a typed knowledge graph, 23 tools.

Rust · MCP · Postgres + pgvector

03

avatar-media-mcp

Generative media over MCP: identity, face swap, voice, frames and assembly for avatar-driven video.

Rust · MCP · ONNX Runtime

04

single-vm-mail-platform

Mail platform as code: Postfix, Dovecot and OpenDKIM on one VM, sized for 500+ mailboxes and 25,000 messages/day.

Terraform + Ansible · 4 vCPU / 8 GB · 114 unit tests

05

single-vm-web-platform

Multi-tenant edge as code: Traefik v3, per-FQDN Compose stacks, shared MySQL, operator plane with forwardAuth.

Traefik · Docker Compose · rebuilds with make bootstrap

01

Security audit and red team engagements

We analyze where your infrastructure breaks: websites, CI/CD pipelines, repositories, secrets, servers, admin panels, and internal tools. Processing runs on a local model by default; your own API keys are supported when you want them. Red Team agents are built under your rules of engagement.

  • Continuous recon
  • Every finding verified by hand
  • Controlled exploitation
  • Purpose-built tooling
02

Private agents on your hardware

We build agents for a specific task. Local models by default, your own keys when you want them. Full engineering cycle: architecture, integrations, interfaces, deployment, and operations.

  • An agent that operates the machine
  • Memory and knowledge stay on this machine
  • Routine runs automated
  • Turnkey infrastructure you own

Client names stay with the client, and engagements stay private. Audit materials are processed on our side on a local model; upon completion, such data is destroyed or returned to you. Agents are developed and deployed on your infrastructure.

// Task

One sentence is enough. We build the system.