AI Engineer

Summary

Lex Sentinel is building a regulated compliance technology platform that helps law firms, wealth managers and estate agents run end-to-end Anti-Money-Laundering (AML), Know-Your-Customer (KYC) and Source-of-Wealth checks with their clients. The platform is a multi-tenant, AI-native SaaS: clients complete identity verification in a secure portal, connect bank accounts via open banking, upload documents and answer AI-generated follow-up questions; the platform reconstructs the wealth narrative, drafts a compliance-grade report and routes it for human sign-off.

We are hiring an AI Engineer to take primary responsibility for shipping production features. You will work directly with the founder on a fixed, modern stack — Lovable, Replit, Supabase, Vertex AI, Claude Code, TypeScript and React — and contribute across the full product surface, from database schema and Row-Level Security policies through Edge Functions and AI integration to frontend UI and audit logging. This is pre-launch, founder-led and well-architected: we are not building prototypes, we are building software that will be reviewed by regulators, auditors and professional indemnity insurers. Salary is competitive, based on experience.

Responsibilities

• Design and ship production features end-to-end on the Lovable/Replit + Supabase + Vertex AI stack — from data model to UI.
• Write idiomatic TypeScript and React; review and correct AI-generated code rather than accept it uncritically.
• Author Supabase migrations, Row-Level Security policies and Edge Functions to a high standard, enforcing tenant isolation at the database layer.
• Integrate AI models (Anthropic Claude, Google Gemini) through Vertex AI's europe-west4 regional endpoint to satisfy EU data-residency commitments, with documented call-site evidence.
• Design, build and evaluate retrieval-augmented generation (RAG) systems end-to-end on Supabase pgvector — embedding pipelines, retrieval strategies, hybrid search, reranking and evaluation.
• Implement and protect the audit-logging layer: immutable, tamper-evident entries with full evidence traceability.
• Use Claude Code, Lovable and Replit productively, with specific, file-anchored, verification-driven prompts — the GitHub repository is the source of truth, never an AI agent's narrative.
• Collaborate on prompt engineering, model evaluation and AI governance; document architectural and compliance decisions so they are defensible to a regulator or auditor.
• Surface security, compliance and architectural risks early.

Requirements

• Minimum two years' post-graduation full-time work experience in the AI field.
• Production experience shipping web applications in TypeScript and React (hooks, react-query, react-hook-form, Zod).
• Hands-on Supabase experience: schema design, migrations, Row-Level Security (you can explain RLS in detail and have written non-trivial policies), Edge Functions and Storage.
• Proven proficiency with AI-native build environments (Lovable and/or Replit) and with Claude Code for hands-on engineering: refactors, multi-file edits, test generation and verification workflows.
• Demonstrable experience designing, building and evaluating RAG systems end-to-end in production — not just prototypes.
• Experience integrating LLM APIs (Anthropic, OpenAI or Google) from a backend: structured output handling, retries, error handling, prompt versioning.
• Sound secret management instincts — you have never put an API key in a browser bundle, and you can explain why.
• Experience in a regulated, audited or compliance-sensitive environment (legal, financial services, healthcare or similar).
• Strong written communication: clear decision records, concise blockers, risks explained without losing substance.
• Comfortable working remotely and asynchronously.

Preferred experience

• Vertex AI (europe-west4 and the Anthropic publisher endpoint pattern); hybrid retrieval (BM25 + vector) and rerankers.
• AI governance: versioned prompt libraries, model evaluation, hallucination guards, structured output validation.
• UK/EU AML/KYC workflows, GDPR, or related regulated SaaS; Plaid, Twilio or ElevenLabs; GitHub Actions CI/CD.

Equal opportunities

Lex Sentinel welcomes applications from candidates of all backgrounds.