Skip to content

About David Sanker · Lawkraft (one-person company)

Lawkraft is me — David Sanker. I’m a lawyer-technologist who designs, builds, and governs AI systems for law, insurance, and finance. No big teams, no hand-offs: you work directly with the founder from scoping to delivery.

Operating brand on this site: AI Innovation — delivered by Lawkraft (Cologne, New York, Brighton).

What I do (in plain language)

  • Strategy with numbers — I map use-cases to value, cost, and risk so you can prioritise confidently.
  • Compliance that passes — EU AI Act classification, DPIA, and the technical file with evidence.
  • Engineering that ships — GraphRAG/RAG pilots, connectors to your DMS/CRM/case systems, and MLOps so it runs reliably.
  • Governance & guardrails — evals, bias/safety tests, audit trails, and handover runbooks.

Typical outcomes (from comparable EU work): faster research/review, quicker claim-triage, measurable cost-per-answer and latency improvements, and no surprises at audit time.

How I work (solo, senior, accountable)

  • Founder-led — I do the analysis, architecture, and core implementation.
  • Tight scope, quick value — Discover (≈2 wks) → Build (≈6–8 wks) → Govern (≈1–2 wks).
  • One or two clients at a time — so you get focus and predictable delivery.
  • EU-first hosting (on-prem supported), mTLS, least-privilege, auditable pipelines.
  • Documentation by default — technical file, DPIA, eval/model cards, runbooks.

If a project needs extra hands (e.g., data labelling or UI polish), I bring in a small trusted network — I remain your single point of contact and code owner.

What I bring (skills & stack)

  • Legal + Engineering: EU tech-law, privacy, model governance and hands-on builds.
  • Retrieval & Reasoning: Knowledge graphs, GraphRAG/RAG, hybrid retrieval (semantic + sparse + graph traversal).
  • MLOps & Ops: CI/CD, evals, prompt/version registries, guardrails, observability (traces/metrics/logs), drift detection.
  • Integrations: DMS, case/matter systems, SharePoint/Confluence, Elasticsearch/OpenSearch; DBs/warehouses.
  • Security: zero-trust, mTLS, secrets management, role-based access, immutable logs.

What I measure: precision@k / recall / MRR, groundedness & citations, TP90 latency, error budgets, cost per answer.

Short background (CV snapshot)

  • Practice: Attorney (Germany), registered NY (foreign legal counsel).
  • Focus: AI/ML for legal and financial services; knowledge graphs; compliant automation.
  • Education: Law (University of Cologne, state exams/clerkship), further study incl. UNICRI and MIT Sloan short programs.
  • Languages: German (native), English (fluent).

Full CV available on request.

Selected work (anonymised)

  • Legal research copilot (GraphRAG) — combined case law, internal memos, and doctrine with citations; materially reduced research time and improved consistency.
  • Prospectus & disclosure automation — templated generation aligned to EU financial regs; orders-of-magnitude unit-cost reduction with human oversight.
  • Claims triage assistant — integrated policy DB + CRM; cut manual routing, improved explanation/appeal handling; added drift & abuse monitoring.

Engagement models

  • Fixed/capped fee — assessments, readiness, compliance accelerators.
  • Time & materials — complex builds with evolving scope.
  • Value-based — when outcomes are quantifiable.

Capacity: I usually book one major build and one advisory track concurrently.

Values (the guardrails I hold myself to)

  1. Sovereignty first — your data stays in your jurisdiction.
  2. Transparency — every model ships with eval cards & bias sheets.
  3. Speed with rigour — quick wins, no shortcuts in compliance.
  4. Auditability — if we can’t explain it, we don’t ship it.

Contact

I reply within 1 business hour.
Email: mail@lawkraft.com · Strategy call: Book 30 min »