Senior Data Scientist

ETHOS - Talent & Advisory; Baltimore (MD); On-site; USD 140,000 - 210,000; Full time

Job summary

ETHOS - Talent & Advisory is seeking a senior data scientist at the intersection of NLP, LLMs, and real-world data challenges. You will design pipelines, shape analytics, and translate complex findings for technical and executive audiences.

You will own entity resolution pipelines, build Python/SQL data processing flows, handle large legacy datasets, and optimize the matching logic with blocking, string distance, phonetics, and probabilistic thresholds.

Qualifications

• One of the following: Master’s degree + 10+ years, Bachelor’s degree + 12+ years, or 18+ years of experience in lieu of a degree.
• Overall 10+ years in the IT industry.
• Strong practical experience with NLP, text processing, and information extraction including NER and address standardization.
• Advanced Python development for analytics solutions and data handling.
• Advanced SQL—complex querying and optimization; experience with data privacy and reproducibility.

Responsibilities

• Design and maintain Python and SQL data processing pipelines on enterprise platforms.
• Clean and transform large-scale datasets from complex legacy sources and tune matching logic (blocking, string distance, phonetic encoding).
• Optimize SQL queries, participate in code reviews, enforce version control and reproducibility, and ensure data privacy.
• Explain algorithmic decisions to executives and non-technical stakeholders for action.

Skills

NLP

LLMs

Python development

SQL

Data processing pipelines

Information extraction

Regex

TF-IDF/cosine similarity

Data privacy

Communication

Education

Master's degree + 10+ years

Bachelor's degree + 12+ years

18+ years experience in lieu of a degree

Statistics

Applied Mathematics

Computer Science

Information Science

Tools

Splink

FastLink

Dedupe

recordlinkage

spaCy

scikit-learn

Job description

This is a hands-on senior role for a data scientist who lives at the intersection of NLP, LLMs, and real-world data problems. You won't just build models you'll design pipelines, shape the analytic approach, and translate complex findings for technical and executive audiences alike. If you're excited about applying Generative AI to meaningful, large-scale challenges, this is worth a look.

WHAT YOU'LL D

OYou'll own the entity resolution pipelines that determine whether two records describe the same person. Day to day, that means designing and maintaining Python and SQL data processing pipelines on enterprise platforms, cleaning and transforming large-scale datasets from complex legacy sources, and tuning the matching logic itself — blocking strategies, string distance metrics, phonetic encoding, and the probabilistic thresholds underneath all of it

The engineering bar is real. You\'ll optimize complex SQL queries and database operations for scale, participate in code reviews, hold the line on version control and reproducibility, and treat data privacy as a requirement rather than a checkbox. You\'ll support validation, testing, deployment, and post-implementation monitoring, so you\'ll see your work through rather than handing it off at the model boundary

The other half of the job is explaining it. You\'ll translate algorithmic decisions — why this match threshold, what a false positive costs here — into terms executive leadership and non-technical stakeholders can act on. Communication is a stated requirement for this role, not a soft preferenc

e.
WHAT YOU N

• EEDOne of the following: a Master\'s degree plus 10+ years of experience, a Bachelor\'s degree plus 12+ years, or 18+ years of experience in lieu of a degree. Degree fields: Statistics, Applied Mathematics, Computer Science, or Information Scien
• ce.Overall 10+ years in the IT indust
• ry.Strong practical experience with NLP, text processing, and information extraction, specifically: – Named Entity Recognition and address standardization on unstructured text – Data matching strategy: blocking and indexing, string distance metrics, phonetic encoding – TF-IDF and cosine similarity for text comparison and information retrie
• valStrong Python development skills for building analytics solutions and manipulating da
• ta.Advanced SQL — complex querying, optimization, and database operatio
• ns.Practical Regex for advanced text processing, data cleansing, and pattern matchi
• ng.Familiarity with the specialized tooling: linkage libraries such as Splink, FastLink, Dedupe, or recordlinkage, plus spaCy for NLP and scikit-learn for machine learning and clusteri
• ng.Comfort with code reviews, version control, and maintaining data security and reproducibility standar
• ds.Excellent written and verbal communication skil
• ls.Ability to obtain and maintain a Public Trust clearan

WHAT MAKES YOU STAND

• OUTPrior delivery of IT or data initiatives in federal, state, or local government environme
• nts.A track record of operating independently — owning a data pipeline architecture end to end, from data discovery through post-implementat
• ion.Experience pulling and migrating data from legacy and distributed systems: PostgreSQL, DB2, Oracle, SQL Server, Hadoop, and unstructured flat fi
• les.Jenkins for CI/CD automation of data validation pipeli
• nes.Pipeline automation tooling for scheduling and monitoring complex data cleansing j
• obs.A demonstrated ability to turn probabilistic match thresholds into business logic an executive can sign off
• on.Strong problem-solving skills and proven verbal and written communication when collaborating across cross-functional te