{
  "updated": "2026-11-06",
  "services": [
    {
      "id": "expert-review",
      "title": "Expert review",
      "audience": "HealthTech, FinTech, LegalTech and enterprise AI teams",
      "summary": "Specialists check model pre-labels on clinical, legal and financial material.",
      "tasks": [
        {
          "task": "Complex reasoning",
          "output": "Step-by-step judgements on multi-part questions, with the reviewer’s correction where the model went wrong."
        },
        {
          "task": "Legal contract breakdown",
          "output": "Clause types, parties, obligations and dates labelled clause by clause."
        },
        {
          "task": "Financial document audits",
          "output": "Line items, figures and entities checked against the source document."
        },
        {
          "task": "Clinical and biomedical indexing",
          "output": "Conditions, drugs, procedures and outcomes tagged against your coding scheme."
        }
      ],
      "url": "https://datanoter.com/services/expert-review/"
    },
    {
      "id": "rlhf-alignment",
      "title": "RLHF and model alignment",
      "audience": "Frontier model labs and open-source fine-tuners",
      "summary": "Response ranking, red-teaming and hallucination checks for language models.",
      "tasks": [
        {
          "task": "Response ranking",
          "output": "Ranked model answers with a short reason for each preference."
        },
        {
          "task": "Adversarial red-teaming",
          "output": "Prompts that break the model, grouped by failure type, with the failing responses."
        },
        {
          "task": "Safety calibration",
          "output": "Labels for refusals that should not happen and answers that should have been refused."
        },
        {
          "task": "Hallucination reduction",
          "output": "Claims in model answers marked as supported, unsupported or wrong, with the evidence."
        }
      ],
      "url": "https://datanoter.com/services/rlhf-alignment/"
    },
    {
      "id": "sensor-fusion",
      "title": "Multimodal sensor fusion",
      "audience": "Autonomous systems, AgTech, robotics and drone startups",
      "summary": "3D LiDAR segmentation and multi-camera tracking, kept in sync across sensors.",
      "tasks": [
        {
          "task": "3D LiDAR segmentation",
          "output": "Point-level classes for each sweep, using your class list."
        },
        {
          "task": "Multi-view camera tracking",
          "output": "Object identities kept consistent across cameras and frames."
        },
        {
          "task": "Sensor-synchronised labelling",
          "output": "One set of labels aligned across LiDAR, cameras and other sensors by timestamp."
        }
      ],
      "url": "https://datanoter.com/services/sensor-fusion/"
    }
  ]
}