{
  "name": "Samuel Abramov",
  "headline": "Freelance Full-Stack Engineer",
  "location": "Hamburg, Deutschland",
  "timezone": "Europe/Berlin",
  "website": "https://abramov-samuel.de",
  "roles": [
    "Freelancer",
    "Full-Stack Developer",
    "Backend Specialist",
    "Machine Learning Practitioner"
  ],
  "availability": [
    {
      "id": "rewe-digital-2026",
      "label": "Jetzt – März 2026",
      "timeframe": "Kontingent belegt",
      "status": "booked",
      "focus": "Full Stack Engineering @ REWE Digital",
      "capacity": "4 Tage/Woche remote",
      "mode": "Remote-first, punktuelle Workshops in Köln",
      "notes": "Langfristiges Mandat, Commerce-Plattform & AI-Features im Fokus."
    },
    {
      "id": "post-rewe-2026",
      "label": "Ab April 2026",
      "timeframe": "Offen für neue Engagements",
      "status": "open",
      "focus": "Greenfield Fullstack Delivery (Java & Angular) sowie AI/ML-Produktentwicklung",
      "capacity": "Vollzeit (4–5 Tage/Woche)",
      "mode": "Remote-first, punktuelle Workshops vor Ort (DACH/EU)",
      "notes": "Bevorzugt Projekte mit klarer Ownership und datengetriebenem Fokus."
    }
  ],
  "preferences": [
    {
      "label": "Arbeitsmodelle",
      "value": "Remote-first, Hybrid in Hamburg oder projektbezogen europaweit vor Ort."
    },
    {
      "label": "Reisebereitschaft",
      "value": "Onsite-Workshops bis zu 2 Tage pro Monat in DACH sind möglich."
    },
    {
      "label": "Response & Onboarding",
      "value": "Reaktionszeit < 24h an Werktagen, Start innerhalb von 3 Wochen realisierbar."
    }
  ],
  "impact": [
    {
      "value": "15+",
      "label": "Jahre Erfahrung"
    },
    {
      "value": "50+",
      "label": "Projekte Delivered"
    }
  ],
  "services": [
    {
      "name": "Backend Engineering",
      "segment": "Enterprise Backends",
      "description": "Skalierbare APIs, Messaging und Datenflüsse für kritische Plattformen mit Fokus auf Zuverlässigkeit.",
      "capabilities": [
        "Microservices & Event-driven Architectures",
        "RESTful & GraphQL APIs",
        "Datenbank-Design & Performance-Tuning",
        "Observability, Security & CI/CD Automatisierung"
      ]
    },
    {
      "name": "Frontend Engineering",
      "segment": "Interfaces mit UX",
      "description": "Responsive Interfaces und Design Systems für komplexe Produktteams mit hoher Delivery-Geschwindigkeit.",
      "capabilities": [
        "Angular, React, Web Components",
        "Design Systems & Component Libraries",
        "Microfrontend-Architekturen",
        "Progressive Web Apps & Performance Audits"
      ]
    },
    {
      "name": "Machine Learning",
      "segment": "Applied AI",
      "description": "Produktionsreife ML-Stacks von der Datenpipeline bis zum Deployment inklusive Monitoring und Governance.",
      "capabilities": [
        "Computer Vision & Predictive Analytics",
        "ML-Ops, Deployment & Model Monitoring",
        "Natural Language Processing",
        "Data Engineering & Feature Stores"
      ]
    }
  ],
  "techStack": [
    "Java",
    "Spring Boot",
    "TypeScript",
    "Angular",
    "React",
    "Svelte",
    "Node.js",
    "Python",
    "TensorFlow",
    "PyTorch",
    "Docker",
    "Kubernetes",
    "AWS",
    "Azure",
    "PostgreSQL",
    "MongoDB",
    "Redis",
    "Kafka",
    "GraphQL"
  ],
  "publications": [
    {
      "title": "AcroMELD: Recovering Interactive PDF Forms with Structure-Aware Graph Set Transformers",
      "authors": [
        "Samuel Abramov"
      ],
      "published": "2026-08",
      "arxivId": "2608.22338",
      "doi": "10.48550/arXiv.2608.22338",
      "subjects": [
        "cs.CV"
      ],
      "summary": "Viele PDFs sehen aus wie Formulare, haben aber keine interaktiven Felder. AcroMELD rekonstruiert sie: Ein Detektor mit 39,4 Mio. Parametern verbindet visuelle Transformer über Graph-Set-Layer mit den PDF-Primitiven und kombiniert visuelle, strukturbasierte und gelernte Recovery-Queries.",
      "url": "https://arxiv.org/abs/2608.22338",
      "pdf": "https://arxiv.org/pdf/2608.22338"
    },
    {
      "title": "Semi-Supervised Learning for Cancer Detection of Lymph Node Metastases",
      "authors": [
        "Amit Kumar Jaiswal",
        "Ivan Panshin",
        "Dimitrij Shulkin",
        "Nagender Aneja",
        "Samuel Abramov"
      ],
      "published": "2019-06",
      "arxivId": "1906.09587",
      "subjects": [
        "cs.CV",
        "cs.AI"
      ],
      "summary": "CNN-Modell auf dem PatchCamelyon-Benchmark (PCam) zur Erkennung von Lymphknotenmetastasen in histopathologischen Scans. Semi-Supervised Training mit Pseudo-Labels verbessert die AUC deutlich gegenüber einer starken CNN-Baseline.",
      "url": "https://arxiv.org/abs/1906.09587",
      "pdf": "https://arxiv.org/pdf/1906.09587"
    }
  ],
  "languages": [
    "Deutsch",
    "Englisch"
  ],
  "engagementModels": [
    "Remote-first",
    "Hybrid (Hamburg)",
    "Workshops europaweit vor Ort"
  ],
  "contact": {
    "email": "info@abramov-samuel.de",
    "phone": "+49 176 5775 3021",
    "linkedin": "https://www.linkedin.com/in/samuel-abramov-54060a121/",
    "github": "https://github.com/Samyssmile"
  },
  "links": {
    "portfolio": "https://abramov-samuel.de/#portfolio",
    "services": "https://abramov-samuel.de/#services",
    "research": "https://abramov-samuel.de/#research",
    "recruiterToolkit": "https://abramov-samuel.de/#recruiter"
  },
  "focusAreas": [
    "Enterprise Engineering",
    "AI & ML Production",
    "Cloud-native Delivery"
  ]
}