Platform Deployment Engineer @ Spacture AI · Brooklyn, New York City

Emre Diricanli

DevOps & Systems Engineer

From bare-metal Kubernetes clusters to AI-native operating systems — deploying edge devices, automating everything, and shipping real systems that work.

View my workGitHubLinkedIn
scroll

01·about

From Izmir to Atlanta to Brooklyn — always building.

I'm a systems engineer who treats infrastructure as a craft. By day I'm at Spacture AI, where I've shipped 50+ end-to-end deployments of AI-powered video analytics — provisioning Linux edge devices with NVIDIA GPUs, wiring up secure Tailscale networks across client sites, and making complex computer-vision systems feel turnkey.

I earned my B.S. in Software Engineering from Kennesaw State University while working full-time and running Alpha Data Tech, the IT and web consultancy I founded. That period taught me the thing I value most: resilience, and the difference between code that demos well and systems that survive production.

Off the clock I run a multi-node Proxmox and Kubernetes homelab, write Go and Rust tooling for networks I administer, and build AI-native systems — including an operating system where Claude is the interface, not an app.

0+

Edge AI deployments shipped

0

Nodes in my homelab cluster

0+

Public repositories

0

Languages in production

izmir → atlanta → brooklyn

02·experience

Real systems, real stakes.

  1. Spacture AI

    Platform Deployment Engineer

    2025 — Present·New York City

    AI-powered video intelligence for retail loss prevention. I own the path from contract to working system.

    • Completed 50+ end-to-end NVR integrations with proprietary AI shoplifting-detection software
    • Provisioned custom Linux edge devices with NVIDIA GPUs — OS installs, drivers, networking
    • Configured secure remote access across 50+ private client networks via Tailscale, static IPs, and port forwarding
    • Delivered hands-on client training and post-onboarding technical support
    LinuxNVIDIATailscaleNetworkingComputer Vision

  2. Alpha Data Tech

    Founder

    2023 — 2024·Atlanta, GA

    Founded an IT solutions and web development company serving small businesses — now my full-stack and DevOps brand.

    • Engineered custom inventory tracking systems that cut client operational costs by ~$200/month
    • Migrated on-premise systems to AWS cloud infrastructure
    • Partnered with small business owners to remove technical bottlenecks and improve workflow efficiency
    AWSWeb DevelopmentIT ConsultingClient Work

  3. Kennesaw State University

    B.S. Software Engineering

    Graduated Dec 2024·Atlanta, GA

    Completed my degree while working full-time — focus on DevOps, cloud computing, and network security.

    • Team Lead for LedgerLink Pro, a full-stack double-entry accounting platform (capstone, delivered on time)
    • Coursework spanning distributed systems, cloud computing, and network security
    Software EngineeringTeam LeadCapstone

03·selected work

Things I've built and actually run.

AI-native operating system

AxiOS

A bootable Linux OS where Claude is the interface, not an app. A Go daemon routes between the Anthropic API and local Ollama models, exposing the entire machine through eight purpose-built MCP servers — filesystem, Docker, GPU, network, and more — under a tiered permission model.

  • Custom Yocto Linux image, flashable to bare metal
  • 8 MCP servers over Unix sockets
  • Hybrid cloud + local AI routing with offline fallback
  • Go
  • Yocto Linux
  • MCP
  • Claude
  • Ollama
  • TypeScript
Source

Local-first AI assistant

E.D.I.T.H.

A self-hosted AI assistant written in Rust with zero cloud APIs. An eleven-tool agent loop handles sandboxed shell and Python execution, self-hosted web search, Obsidian vault integration, and persistent SQLite memory — plus a Python voice pipeline with wake word, STT, and TTS.

  • Agent + tool-calling architecture from scratch in Rust
  • Entirely local inference via Ollama
  • Long-term memory and knowledge-base integration
  • Rust
  • tokio
  • SQLite
  • Ollama
  • Python
Private build · GitHub

AI paper-trading terminal

ai_trader

A dense, btop-style trading TUI where FinBERT scores market news and a local LLM generates signals. Automated execution is double-gated: trades fire only on confidence thresholds and backtest-certified strategies, with a global kill switch.

  • Python
  • Textual
  • PyTorch
  • Ollama
  • Alpaca API

Bare-metal Kubernetes homelab

AlphaCluster

A multi-node Proxmox and k3s cluster on HP workstations, run like production: Ansible-automated provisioning, Prometheus and Grafana observability, Vault-managed secrets, and a Tailscale-exposed API so the cluster is reachable from anywhere.

  • Proxmox
  • Kubernetes
  • Ansible
  • Prometheus
  • Grafana
  • Vault

Network discovery CLI

ndscan

A modern Go wrapper around nmap that makes daily network discovery fast: clean table and tree views, MAC vendor lookup, JSON export for pipelines, and remote scanning through SSH jump hosts with plain user@host syntax.

  • Go
  • Cobra
  • nmap
  • SSH

Live tool · recordingcalculator.com

Retention Calculator

A client-side NVR storage calculator that models per-stream bitrate from resolution, FPS, codec, and quality. Its constants are calibrated against real recordings from 24 production cameras across 5 stores — and deliberately tuned conservative.

  • TypeScript
  • Next.js
  • React
  • Tailwind CSS
see everything on github

04·stack

Tools I reach for.

Infrastructure & DevOps

  • Kubernetes
  • Docker
  • Proxmox
  • Ansible
  • Terraform
  • AWS
  • GCP
  • CI/CD

Languages

  • Go
  • TypeScript
  • Python
  • Rust
  • Java
  • Bash

Networking & Security

  • Tailscale
  • Twingate
  • nmap
  • VPN & DNS
  • HashiCorp Vault
  • Pi-hole

Observability

  • Prometheus
  • Grafana
  • Uptime Kuma
  • SLO Dashboards

Web & Product

  • Next.js
  • React
  • Tailwind CSS
  • Node.js
  • PostgreSQL
  • Supabase

AI Systems

  • Claude API
  • MCP
  • Ollama
  • Local-first LLMs
  • Agent Architectures

05·beyond the terminal

What I do when the cluster is green.

Photography

Street, portrait, and night photography on 35mm and 85mm primes. RAW workflow, Lightroom, and a zine someday.

Honda S2000

My garage project. Injectors, fluids, and alignment done — suspension refresh is next on the lift.

Formula 1

Race weekends are sacred. Made the pilgrimage to COTA — camera in hand, of course.

Homelab

Four humming nodes in Brooklyn. If it can be self-hosted, automated, or monitored, it already is.

06·contact

Let's build something that survives production.

I'm always open to interesting problems — infrastructure, AI systems, or something nobody has a name for yet. The fastest way to reach me:

I'll get back to you within 24 hours.