AI Evaluation Engineer

Benchmark & Evaluation Engineer

Type: Full-time Location: Remote (Philippines) Schedule: US Shift

We are a tech-enabled workforce augmentation platform leveraging top 1% offshore talent & cutting edge technology to enable businesses to unlock their full potential.

Our client is an innovative AI platform company developing industry-standard benchmarks to evaluate how AI analysts handle complex, production-scale enterprise data. They are remote-first, fast-growing, and building products that bridge natural language analytics, data modeling, and end-to-end data systems.

We are looking for a Benchmark & Evaluation Engineer with a strong mix of data engineering, advanced SQL, and AI evaluation experience. You should have a "production data" sensibility—knowing how complex, messy, and multi-system enterprise environments operate—and a bias for rigor and honest scoring.

What You'll Do

• Build Realistic Data Environments: Design and stand up multi-system enterprise environments (warehouses, lakes, APIs, DBs) at scale across key domains like healthcare, finance, and product analytics.
• Engineer Benchmark Datasets: Generate calibrated, large-scale datasets with realistic noise, seasonality, drift, and synthetic PII to test query scalability and reasoning.
• Curate Task Libraries: Write and verify benchmark tasks, golden SQL, and rubric-scored reasoning keys while eliminating data leakage and ambiguity.
• Own the Evaluation Harness: Extend grading systems (deterministic checks, LLM judges), maintain regression-tracked leaderboards, and ensure scoring integrity.

What You Need

Non-Negotiable Qualifications:

• Advanced SQL & Data Modeling: Expert proficiency in production-grade SQL (CTEs, window functions, query plans, cardinality) and data modeling (star/snowflake schemas, SCDs, referential integrity).
• Data Engineering & Systems: Strong Python skills (pandas, NumPy, Polars) and hands-on experience standing up and loading data into at least one major warehouse (Snowflake, BigQuery, Redshift, or Postgres).
• Evals & Benchmark Experience: Proven experience designing evaluation sets or benchmark tasks, including golden-answer/golden-SQL verification, rubric/LLM-judge calibration, avoiding data leakage, and tracking regressions.
• Multi-System Data Environments: Ability to construct realistic multi-system setups (warehouses, data lakes, operational DBs, APIs) with large-scale, semi-structured, or dirty synthetic data.

Nice to Have:

• Domain knowledge in Healthcare/Health Insurance, Finance/FP&A, Product Analytics, or Supply Chain.
• Experience with LLM evaluation frameworks (e.g., SWE-bench, Cursor-style harnesses) or streaming/data-lake stacks (S3, Parquet, dbt, Airflow, Spark, DuckDB).

Candidate Data & Privacy Notice

By submitting your application to Expedock, you acknowledge and consent to the collection, use, and processing of your personal information for recruitment and hiring purposes. Your information will be used to:

• Evaluate your qualifications and suitability for current and future roles
• Communicate with you throughout the recruitment process
• Improve our hiring processes and overall candidate experience
• Maintain talent pools for future opportunities, where permitted by law

We handle candidate data with care and in accordance with applicable data protection and privacy regulations. Your information will only be accessed by authorized team members and will not be shared with third parties without your consent, unless required by law.