Senior Data Scientist — LLM/NLP for Census Insights

Ignite IT, LLC; Suitland, Northern (MD, KY); Hybrid; USD 120,000 - 180,000; Full time

Benefits offered by this job

401(k)

401(k) matching

Dental insurance

Flexible spending account

Health insurance

Life insurance

Paid time off

Professional development assistance

Referral program

Tuition reimbursement

Vision insurance

Job summary

Ignite IT, LLC seeks a Senior Data Scientist to improve the Census Bureau’s LLMAutocoder and evaluate its use across household surveys and demographic projects. You will lead analytical methods and experimental design for coding survey write-in responses, working with engineers, the Model Evaluation Analyst, and Government SMEs to turn evidence into actionable recommendations.

You will establish baselines for ACS Autocoder, analyze coding patterns and error rates, and design reproducible

Qualifications

• Bachelor's degree or equivalent professional experience required.
• 8+ years in applied data science or statistics, incl. 3+ years NLP/LLM.
• Strong Python/SQL skills and large dataset experience.
• Solid experimental design and model evaluation skills.

Responsibilities

• Establish baselines and evaluation methods for ACS Autocoder.
• Analyze coding accuracy, rates, coverage, and error patterns for improvement opportunities.
• Design reproducible experiments with representative samples and controls.
• Compare LLMs, fuzzy matching, and dictionary-based approaches for coding quality and cost.
• Assess data readiness and coding taxonomies for additional surveys and ACS fields.
• Collaborate with Government SMEs to refine prompts, preprocessing, and model configuration.
• Interpret results and document findings and guidance for adoption or revision.

Skills

Python

SQL

NLP

LLM applications

Experiment design

Communication

Education

Bachelor's degree in data science, statistics, computer science, mathematics, or related field

Advanced degree preferred

Tools

AWS

Job description

Ignite IT, LLC seeks a Senior Data Scientist to improve the Census Bureau’s LLMAutocoder and evaluate its use across household surveys and demographic projects. You will lead analytical methods and experimental design for coding survey write-in responses, working with engineers, the Model Evaluation Analyst, and Government SMEs to turn evidence into actionable recommendations.

You will establish baselines for ACS Autocoder, analyze coding patterns and error rates, and design reproducible