AI & Data Intelligence Engineer

Role Overview
We are seeking a specialized AI & Data Intelligence Engineer to lead the accuracy, prompt architecture, and statistical methodology engine for EquiSight—our enterprise solution for job architecture & design, compensation studio, and pay equity compliance.
Working alongside our data architect, development team, and domain experts, you will directly manage EquiSight’s language model pipelines structured schema enforcement, versioned prompt templates, and Python statistical algorithms. The sole primary responsibility is ensuring that Equisight’s AI workflows and statistical models perform deterministically, resist drift, and meet strict regulatory standards.
Key Responsibilities
1. LLM Engineering & Gateway Guardrails:
Author, maintain, enhance, and version control EquiSight’s prompt templates.
Enforce zero-hallucination policies using strict JSON Schema validation and Pydantic models. Ensure low-confidence or failed outputs trigger automated retries or human-in-the-loop review routes.
Manage the AI Gateway module to log token usage, model parameters, confidence scores, and verbatim evidence citations.
2. Statistical Analytics Engine & Verification:
Co-own and update the pure-Python analytics engine executing the primary metrics and analytics engine in Equisight with the data architect.
Build and maintain statistical validation test harnesses that compare/backtest production outputs against reference datasets across the industry and market data.
3. Compliance, Governance & AI Act Integrity:
Enforce a strict architectural boundary: AI performs natural language parsing/drafting only, while deterministic code executes all arithmetic, scoring, and legal threshold checks.
Support high-risk auditability expectations by pinning methodology versions, prompt templates, and model parameters directly in Git source control.
Explicit Technology Stack Breakdown
1. Technologies Owned & Managed by the AI Engineer:
Hosted AI Inference: Anthropic Claude (claude-opus-5, claude-sonnet-5) deployed via Microsoft Foundry (Azure EU Region).
Prompting & Schema Validation: JSON Schema, Pydantic, Jinja2 template engines.
Statistical & Scientific Python: Python 3.11+, statsmodels, scipy, numpy, pandas (for OLS regression, Gelbach decomposition, and robust covariance estimation).
Testing & Model Evals: pytest, automated statistical diff harnesses, LLM evaluation tools (e.g., DeepEval, Promptfoo).
Version Control: Git / GitHub for prompt and methodology versioning alongside release branches.
2. Shared Application Touchpoints:
Worker Execution: Writing task payloads for Celery workers and Redis queues.
Data Layer: Querying workspace schema tables in PostgreSQL via Python/ORM.
3. Technologies Handled by External Partners (Out of Scope for this Role):
Full-Stack Web Development: Django 5, Django REST Framework, Vue 3, Vite.
Cloud & Web Infrastructure: nginx, automatic TLS, Azure Virtual Networks, Azure Blob Storage provisioning, Docker container orchestration, deployment pipelines via GitHub Actions.
Qualifications & Skills
Required:
Education: Bachelor’s or Master’s degree in Data Science, Computer Science, AI/ML, Quantitative Economics, or a related quantitative/technical field.
LLM Prompt Engineering: Hands-on experience prompting and orchestrating frontier LLMs (specifically Anthropic Claude via Azure / Microsoft Foundry).
Schema Validation & Guardrails: Advanced skill in structured output enforcement using JSON Schema and Pydantic.
Advanced Python & Statistics: Strong background in statistical programming with pandas, statsmodels, and scipy for running OLS regressions, log transformations, and error adjustments.
Automated Testing & Evals: Proven track record building evaluation harnesses (pytest, custom eval suites) to measure model drift, citation accuracy, and hallucination resistance.
Preferred (Nice to Have):
HR Domain Knowledge: Background in compensation analytics, job evaluation frameworks, pay equity modeling, or workforce analytics.
Regulatory Compliance: Familiarity with the EU AI Act (Regulation 2024/1689) or the EU Pay Transparency Directive (2023/970) and similar legislations in other key-market regions (Asia, Americas)

Work arrangement
Yes