DevOps/SRE Engineer with 3+ years of experience building fault-tolerant infrastructure and automating CI/CD processes. Achieved 99.9% uptime over 6 months with Proxmox VE HA cluster. Practical experience with ML/Computer Vision infrastructure (Frigate, YOLO, face recognition). SRE approach: runbooks, RTO/RPO metrics, monitoring, automation-first mindset. Reduced operational toil from 40% to 10% through systematic automation.

Infrastructure: Linux (Debian/Ubuntu), Proxmox VE/PBS HA cluster (3 nodes), Docker, KVM/LXC, HP iLO
Automation: Bash, systemd, GitHub Actions, CI/CD
Databases: PostgreSQL (replication, backups)
Networking: nginx (reverse-proxy, TLS termination), VLAN segmentation, DHCP, DNS (AdGuard Home), KeenDNS, Let's Encrypt, NTP, iptables/firewall
Storage: ZFS, Ceph
Monitoring: Dashboards (CPU/RAM/Disk/GPU), alerting (Telegram Bot), centralized logging, capacity planning
ML/Computer Vision: Frigate, YOLO, Doubletake, Compreface, Docker-based model serving, MQTT integration
API & Integrations: REST API, OAuth2, Modbus/TCP, Zigbee2MQTT, Telegram Bot API
Security: SSL/TLS automation, secrets management, firewall policies

Hands-on practice in lab environment: Kubernetes/Helm, Terraform, Ansible, Prometheus/Grafana, GitLab CI, Python (Flask/FastAPI)

OAuth2 infrastructure automation for HeadHunter API:

• ✅ CI/CD via GitHub Actions: ShellCheck linting, nginx validation, Docker build/push to GHCR
• ✅ Automated token refresh via systemd timers (every 6 hours, 100% automation)
• ✅ Reverse-proxy setup: nginx + KeenDNS + Let's Encrypt (90-day cert auto-renewal)
• ✅ Infrastructure as Code: all configs version-controlled in Git
• ✅ Container versioning: SemVer tagging (1.0.0, latest) published to GHCR
• ✅ Architecture documentation: Mermaid diagrams with detailed component interaction

Production-ready ML infrastructure for real-time video analytics:

• ✅ Frigate + YOLO object detection in Docker (8+ video streams, 99.9% uptime)
• ✅ Doubletake + Compreface face recognition (trained on 150+ photos, accuracy 98%, recall 95%)
• ✅ ML optimization: precision improved from 75% to 92%, false positives reduced by 85%
• ✅ Automated model serving via Docker + Python + REST API + MQTT
• ✅ GPU monitoring and inference metrics tracking
• ✅ Full ML lifecycle implementation: data preparation → training → serving → monitoring → optimization

Technologies: Frigate, YOLO, Doubletake, Compreface, FaceNet, MTCNN, Docker, MQTT, Python, REST API, TensorFlow Lite, GPU inference

• ✅ Reduced deployment time by 96%: 2 days → 2 hours via Docker containerization and CI/CD
• ✅ Reduced operational toil by 75%: from 40% → 10% through automation of routine tasks
• ✅ 99.9% infrastructure uptime over the last 6 months (Proxmox VE HA cluster, 3 nodes)
• ✅ RTO < 30 minutes for critical VM recovery (verified via quarterly test-restores)
• ✅ RPO < 1 hour via snapshot management and incremental backups
• ✅ Network load reduced by 15% through AdGuard Home DNS filtering
• ✅ ML model optimization: precision 75% → 92%, false positives reduced by 85%
• ✅ CI/CD pipeline: ShellCheck (100% compliance) + automated Docker builds to GHCR
• ✅ 100% automation of OAuth2 token refresh (manual every 2 weeks → systemd timer every 6 hours)

Proxmox VE HA Cluster:

• 3-node cluster with automatic failover and quorum configuration
• 15+ virtual machines (Ubuntu/Debian)
• 20+ LXC containers for service isolation
• ZFS/Ceph storage backend
• Daily incremental + weekly full backups via Proxmox Backup Server
• HP iLO remote management for physical servers

Network Architecture:

• 10+ VLAN segments (management, production, DMZ, guest)
• 50+ managed devices with DHCP reservations and MAC binding
• nginx reverse-proxy with TLS termination
• AdGuard Home DNS filtering (15% network load reduction)
• NTP synchronization (<10ms accuracy) for distributed logging

Observability:

• Custom dashboards: CPU/RAM/Disk/GPU utilization, network metrics, service health
• Telegram alerting: CPU > 80%, Disk > 90%, service downtime
• Centralized logging with 30-day hot / 90-day cold retention
• Capacity planning based on trend analysis

flowchart TB
    subgraph infra["📦 OAuth2 Infrastructure (This repository)"]
        direction TB
        Nginx[🔄 nginx<br/>HTTP:80→:8000]
        
        subgraph automation["⚙️ Automation"]
            direction LR
            Timer[⏱️ systemd<br/>Every 6h]
            Script[📜 refresh.sh<br/>Token refresh]
        end
        
        TestServer[🧪 test-8000.py<br/>Test server]
        TokenStore[(🔐 Token Storage<br/>/var/lib/hh-token/token.json)]
    end
    
    subgraph prod["🤖 Production App (Separate project)"]
        direction TB
        TelegramBot[📱 Telegram Bot<br/>HH job search]
        FlaskApp[🌐 Flask Application<br/>OAuth callback handler at /callback]
        
        TelegramBot -.->|Project<br/>hh-oauth2-keendns-nginx-systemd| FlaskApp
    end
    
    subgraph external["External"]
        HHAPI[🏢 HH OAuth2 API<br/>api.hh.ru]
    end
    
    %% Connections / Component interactions
    
    %% Main Flow (OAuth):
    Nginx -->|1. Proxy :8000| FlaskApp
    FlaskApp -->|2. OAuth callback| HHAPI
    FlaskApp -->|3. Saves initial tokens| TokenStore
    
    %% Production Flow:
    FlaskApp -->|4. Reads tokens| TokenStore
    TelegramBot <-->|5. API requests| HHAPI
    
    %% Token Refresh Flow:
    Timer -.->|6. Trigger| Script
    Script -->|7. Refreshes tokens| HHAPI
    Script -->|8. Saves new tokens| TokenStore  
    
    %% Testing:
    TestServer -->|9. Testing alternative| Nginx
    
    %% Styling
    style Nginx fill:#2E8B57,color:#FFFFFF,stroke:#1a5f3a,stroke-width:2px
    style Timer fill:#FFA500,color:#000000,stroke:#cc8400,stroke-width:2px
    style Script fill:#FF8C00,color:#FFFFFF,stroke:#cc7000,stroke-width:2px
    style TestServer fill:#DAA520,color:#000000,stroke:#b8860b,stroke-width:2px
    style TokenStore fill:#9370DB,color:#FFFFFF,stroke:#6a4db8,stroke-width:2px
    
    style TelegramBot fill:#4682B4,color:#FFFFFF,stroke:#1565c0,stroke-width:2px
    style FlaskApp fill:#4169E1,color:#FFFFFF,stroke:#2a4ba8,stroke-width:2px
    
    style HHAPI fill:#DC143C,color:#FFFFFF,stroke:#a00000,stroke-width:2px

1. hh-oauth2-keendns-nginx-systemd
   Automated OAuth2 infrastructure for HeadHunter API with reverse-proxy nginx and systemd timers

2. do6pbln9l
   DevOps/SRE Engineer | Linux, Docker, CI/CD | 🏠 Remote

💼 HH.ru: DevOps/SRE Resume

💬 Preferred contact method: Apply via HH or email from resume

🏠 Working from home | 🌟 Open to DevOps/SRE opportunities