Senior Data Scientist (NLP and Unstructured Data Analytics)

Node.Digital•Washington, DC

5d•Remote

About The Position

We are seeking a Senior Data Scientist with expertise in Natural Language Processing (NLP) and Unstructured Data Analytics to join our team. This role involves integrating and scaling NLP methods to analyze large volumes of text data, designing and implementing statistical and machine learning models for financial fraud detection, and collaborating with criminal investigators. You will be responsible for data quality analysis, developing case leads, documenting methodologies for evidentiary requirements, and creating visualizations and dashboards. This position requires close coordination with data engineering to ensure efficient pipelines and the creation of programming and automation techniques using various tools. The goal is to identify new business questions and expand the scope of analysis and reporting.

Requirements

Master's, Ph.D., or doctorate level equivalent degree in data science, machine learning, computer science, mathematics, or a related field. Alternatively, ten years of applied work experience in any of the same fields.
5+ years Designing, implementing, and maintaining advanced AI systems and predictive models, including both supervised and unsupervised models.
5+ years Developing analytic rules and models using leading edge analytic tools and best practices.
5+ years Developing regression, classification, and other statistical models to identify anomalies, patterns, and predictive variables.
3+ years Providing data support for criminal investigations into financial fraud or abuse of government funds.
3+ years Manipulating data in Python. Pandas is required.
3+ years Working in a modern cloud environment: Azure, AWS, or GCP.
2+ years Conducting advanced data analysis in SQL, specifically SQL Server and PostgreSQL.
2+ years Developing and scaling natural language processing solutions.
2+ years Presenting methods and findings to technical and non technical stakeholders, both orally and in written products and visualizations.
Must have a Public Trust Clearance.

Nice To Haves

Production experience with named entity recognition and entity resolution across messy document corpora.
Retrieval augmented generation, vector stores, embeddings, and semantic search at scale.
Large language model integration under federal security constraints, including boundary controlled deployment and prompt versioning.
Optical character recognition pipelines applied to scanned or low quality source documents.
Topic modeling, document classification, or clustering applied to audit, legal, or investigative text.
Cloud certification in Azure, AWS, or GCP.

Responsibilities

Integrate and scale natural language processing methods to parse, clean, and analyze large corpora of unstructured and semi structured text, using optical character recognition, semantic similarity algorithms, and large language models as needed.
Design, develop, test, calibrate, and implement statistical and machine learning models targeting financial fraud, improper payments, and non compliance within SBA programs.
Build and refine supervised and unsupervised models, including regression, Bayesian, clustering, and ensemble approaches.
Review, maintain, and support all existing loan fraud indicators developed by TSD.
Perform data quality analysis on source tables and develop repeatable processes for combining and analyzing large data sources.
Collaborate directly with criminal investigators to determine and execute analytic strategies supporting loan fraud cases, and adhere closely to the federal rules of criminal procedure governing protected information, including Rule 6(e).
Develop case leads for SBA OIG investigations from model outcomes.
Document all methodology, test models, and production models in a form that satisfies criminal evidentiary requirements.
Build visualizations and dashboards conveying methodological choices, outcomes, and predictive capability, iterated on end user feedback.
Deliver findings in multiple registers: data summaries and visualizations for investigative staff, executive summaries for OIG leadership.
Coordinate with the data engineering seat so the architecture supports machine learning and text processing pipelines efficiently.
Create programming and automation techniques using SharePoint, Python, Excel, Power BI, Power Apps, and similar tools.
Identify new business questions that expand the scope of analysis and reporting.