2 months ago
USA (city not specified)Mid Level
H1B sponsor
Base Salary
$215k - $270k/yr
Responsibilities
- Design and build grounding systems connecting agent reasoning to verified enterprise data sources.
- Build and optimize RAG, hybrid search, and structured query generation pipelines for accuracy and latency.
- Define data representation strategies for catalogs, schemas, lineage, and other heterogeneous enterprise data.
- Create automated benchmarks, regression suites, and human evaluation protocols.
- Convert validated research findings into production systems.
- Establish quality metrics and dashboards tracking agent correctness over time.
- Build feedback loops that turn user interaction data into evaluation datasets and grounding improvements.
- Own experiments from hypothesis through production deployment.
Requirements
- At least 3 years of experience in information retrieval, NLP, knowledge representation, or applied ML research.
- Production experience building RAG, grounding, or retrieval systems rather than only prototypes or demos.
- Strong evaluation methodology, including benchmark design, statistical analysis, and reproducible experiments.
- Ability to operate at the research/systems boundary by reading papers and shipping production code.
- Fluency in Python and experience with vector databases, embedding models, and LLM APIs.
- Track record of converting research insights into shipped production systems.
- Preferred: experience with enterprise data systems, SQL engines, data catalogs, schema metadata, text-to-SQL, structured query generation, evaluation pipelines, JVM-based systems, or published research/open-source contributions in IR, NLP, or evaluation methodology.
Benefits
- Equity packages in the form of ISOs.
- Flexible paid time off.
- Comprehensive benefits offering.
- Approximately 25% in-person travel is required for onboarding, offsites, customer engagements, and company events.
- Competitive pay and stock grants.
Categories
About Starburst
Starburst builds an enterprise data platform that lets teams query and govern distributed data across clouds and on‑prem systems without copying it, powered by Trino and Apache Iceberg. It sells commercial Trino-based software and a managed cloud service for federated SQL, data lakehouse analytics, and governance. Founded in 2017 and headquartered in Boston, the privately held company serves global enterprises; customers include Comcast and Citigroup.