Data Associate (AI Labelling & Evaluation)

Arva AI
London
Workplace: OnsiteFull timeGBP 30,000 - 45,000 annuallyFunction: Data Science & Machine LearningExperience: 1-2 yearsSkills: ["Attention to detail","Judgement","Ownership","Written communication","Curiosity"]

Build the ground-truth dataset that trains and evaluates AI agents for financial crime detection. Label KYB/KYC cases, screening hits, sanctions/PEP/adverse media, and transaction monitoring alerts, then grade agent outputs against gold-standard answers and track quality metrics across releases. Define and refine labeling guidelines, taxonomies, and edge-case playbooks, discover ambiguous cases for evaluation sets, and maintain clean, versioned, auditable data in an early-stage compliance+AI environment.

Loading

Loading job details...

Preparing the role view and application actions.

FursaFursa
Arva AI
Arva AI
19 hours ago

Data Associate (AI Labelling & Evaluation)

✓ Verified Job

Canonical indexed version, validated from employer's careers page.

Source: Company careers pageValidated by: Fursa AI
Last checked: 8 hours agoStatus: Live

Job Summary

Build the ground-truth dataset that trains and evaluates AI agents for financial crime detection. Label KYB/KYC cases, screening hits, sanctions/PEP/adverse media, and transaction monitoring alerts, then grade agent outputs against gold-standard answers and track quality metrics across releases. Define and refine labeling guidelines, taxonomies, and edge-case playbooks, discover ambiguous cases for evaluation sets, and maintain clean, versioned, auditable data in an early-stage compliance+AI environment.
Location: London
Workplace: Onsite
Employment Type: Full time
Job Function: Data Science & Machine Learning
Seniority: Entry level

Key Responsibilities

  • •Own labelling and annotation of financial crime data (KYB/KYC cases, screening hits, transaction alerts) used for training and evaluation.
  • •Define what “correct” means for agent decisions and hold model releases to that standard as ground truth.
  • •Grade AI Agent outputs against gold-standard answers, flag errors/inconsistencies, and track quality metrics across releases.
  • •Help define and refine labelling guidelines, decision taxonomies, and edge-case playbooks to keep labels consistent.
  • •Surface ambiguous/novel/adversarial cases from live data, convert them into evaluation sets, and communicate error patterns to Engineering and Product.

Pay and Benefits

Salary: GBP 30,000 - 45,000 annually
Equity and Bonus:Equity

Key Requirements

  • •1–2 years of professional experience in the workplace.
  • •Early-career candidate with strong attention to detail and genuine interest in AI (e.g., law, finance, criminology, data, linguistics).
  • •Ability to learn quickly; no compliance experience required.
  • •Obsessive attention to detail and consistency through high volumes of consequential review work.
  • •Comfort making defensible calls on ambiguous cases and documenting reasoning clearly.
Experience:1-2 yearsFinancial crimeAMLKYCComplianceFintechAI
Skills:Attention to detailJudgementOwnershipWritten communicationCuriosity
Tech Stack:SpreadsheetsAnnotation toolsSQLPython

Eligibility

Nationality:US National

Company Brief

Arva AI
Develops AI-driven perception and autonomy software for commercial and industrial vehicles, enabling safe, efficient self-driving operations using sensor fusion, computer vision, and machine learning to automate logistics, last-mile delivery, and facility transport.
Industry: Autonomous Vehicles
Website