Shawna Staff
Full-stack engineer, end-to-end.
Patent-pending graph IAM · production AI/LLM systems · cloud infrastructure. From database to deployment, I own the whole stack — and I ship.
About Me
Full-stack engineer who owns the whole stack — from database schema to production deployment. My work spans patent-pending graph IAM, production AI/LLM systems, and cloud infrastructure.
I've shipped a multi-tenant fintech compliance platform, have a patent-pending approach to graph-based identity management, run a 30+ service Docker homelab, and built a self-healing AWS compute fleet for original physics research. Backgrounds in Go, Python, React / React Native, and .NET.
Skills
> Languages
> Frontend
> Backend
> Data
> Cloud & DevOps
> AI / ML
Projects
> IAM-GO / Nexus
Graph-based identity and access management with cryptographically assured credential linking (patent pending). Multi-hop Neo4j traversal, ML-based progressive trust scoring, full SoD conflict detection, and a built-in BI engine covering market mapping, influence analysis, and M&A due diligence. OAuth, SAML, OIDC, Auth0, and Firebase auth — all in Go.
> Buckler Due Diligence
Investment analytics and due diligence platform built for Buckler AI. Enables investment professionals to screen and analyze securities, run peer comparisons, perform structured due diligence workflows, and export research — all from a single dashboard. FastAPI backend with HTMX-powered frontend and Chart.js visualization.
> Cascade Inversion Cosmology
A computational research project modeling cosmic structure as self-organized criticality in a structured substrate. 1D Oslo and 2D/3D Manna sandpile simulations reproduce published critical exponents (τ≈1.55, D≈2.2) through finite-size scaling across lattice sizes — validation against the physics literature, not just plausible output. The scaling sweeps run on a self-healing AWS spot fleet: one EC2 instance per job, with spot-interruption auto-recovery, checkpoint-resume, and rsync fleet sync. Python, NumPy/SciPy, and a rigorous, documented methodology.
> Homelab — Production Infrastructure
31+ production services across 14 Docker Compose stacks on a retired Mac. Core stack: Nginx, PHP-FPM, MySQL, PostgreSQL/TimescaleDB, MongoDB, Redis. Networking: Pi-hole, WireGuard VPN, Nginx Proxy Manager, Portainer. Monitoring: Prometheus, Grafana, cAdvisor, node-exporter. Media: Jellyfin, Nextcloud. Home automation: Home Assistant, Mosquitto MQTT, Node-RED. Jenkins CI/CD health-checks all 36 containers. AI agents run daily briefings, security audits, and backup verification automatically.
> Cipher
An AI coding assistant that doesn't forget. Blockchain-backed persistent memory means context is cryptographically verifiable across sessions — not just cached. A knowledge graph structures what it learns, and QLoRA fine-tuning lets the model actually improve from experience. Built for engineers who are tired of re-explaining their codebase every session.
> Dead Life Inc v2
Multi-tenant platform for bands to sell merch, stream live shows, and manage their entire fan economy. Concert ticketing (including private concert access), print-on-demand merchandise, loyalty rewards, real-time analytics, blog, and revenue sharing — each band operates as a fully isolated tenant. 65+ granular permissions across 4 roles, Firebase Cloud Functions backend.