Data Scientist
Bethesda, MD
On-site
Job details
- Bethesda, MD
- On-site
- Full-time
- Platform & DevOps
- Posted 1 month ago
Sponsorship history
Past filings by Black Canyon Consulting in public government data. Not a promise of sponsorship.
- Filed 6 H-1B LCAs in FY2025
Full job description
Overview
We are seeking a Data Scientist to support applied machine learning, natural language processing, AI evaluation, data quality, and analytical product development for a federal permitting modernization initiative.
The Data Scientist will work across structured and unstructured data, including public comments, permitting records, compliance information, documents, and geospatially linked datasets. A key focus of the role will be developing representative evaluation datasets and quantitative methods to measure the performance, reliability, and limitations of AI-enabled and agentic solutions.
This position is remote.
Duties & Responsibilities
- Develop analytical and machine-learning approaches supporting permitting and natural-resource use cases.
- Build and maintain labeled datasets, evaluation corpora, and benchmark datasets for AI-enabled services.
- Define quantitative evaluation methodologies for classification, extraction, retrieval, summarization, entity matching, and other AI/NLP capabilities.
- Evaluate LLM and agentic workflows, including output quality, retrieval performance, task completion, failure modes, and reliability.
- Analyze public-comment datasets and develop reusable features and evaluation datasets for comment-processing services.
- Apply NLP and machine-learning techniques to structured and unstructured data.
- Evaluate embedding models, semantic retrieval strategies, document-ranking techniques, and AI-generated outputs.
- Develop statistical approaches for detecting anomalies, reconciliation issues, missing data, and inconsistent records.
- Partner with Content Pipeline Engineers to measure OCR, extraction, and document-retrieval performance.
- Partner with Data Engineers to ensure analytical datasets and evaluation pipelines are reproducible and production-ready.
- Support entity matching and harmonization across data originating in multiple source systems.
- Design experiments and validation approaches with clearly defined success metrics.
- Contribute to certified analytical data products, metrics, and governed datasets.
- Develop reproducible Python analyses, models, evaluation pipelines, and technical documentation.
- Communicate model performance, limitations, uncertainty, and failure modes to technical and non-technical stakeholders.
Educational Requirements
- Bachelor's degree in Data Science, Statistics, Computer Science, Mathematics, Engineering, or a related quantitative discipline.
