Ignite IT Hub is seeking a Senior Data Scientist to improve the U.S. Census Bureau's LLM Autocoder and assess 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 subject matter experts (SMEs) to turn evidence into practical recommendations.
Responsibilities
- Establish performance baselines and evaluation methods for the American Community Survey (ACS) Autocoder and proposed expansion efforts.
- Analyze accuracy, industry and occupation joint coding rates, coverage, error patterns, and category-level results to identify improvement opportunities and regressions.
- Design reproducible experiments with representative samples, appropriate comparison groups, and controls against data leakage; explain uncertainty and limitations.
- Compare LLMs, fuzzy matching, direct-match dictionaries, rules, and hybrid approaches based on coding quality, explainability, feasibility, and operational cost.
- Assess data readiness and coding taxonomies for additional surveys and ACS fields, such as race and ancestry, field of degree, and language.
- Work with Government SMEs to resolve ambiguous cases and refine evaluation criteria; guide engineers on changes to prompts, preprocessing, model configuration, and coding methods.
- Interpret pilot and production results, recommend whether to implement or revise a method, and document findings, reusable practices, and technical guidance.
Requirements
Required qualifications
- Bachelor's degree in data science, statistics, computer science, mathematics, or a related field, or equivalent relevant professional experience.
- 8+ years in applied data science or statistical analysis, including 3+ years with NLP or text classification and hands-on LLM application experience.
- Strong Python and SQL skills, with experience preparing and analyzing large datasets and using statistical and machine learning libraries.
- Strong experimental design and model evaluation skills, including sampling, class imbalance, error analysis, and reliable comparisons against baselines.
- Experience supporting deployed models, collaborating with engineers, and explaining analytical findings and limitations to technical and nontechnical audiences.
Preferred qualifications
- Advanced degree in a relevant field; experience with survey data, industry and occupation coding, or other detailed classification taxonomies.
- Experience with AWS-based analytical workflows and federal statistical or sensitive-data environments.
Conditions of employment. Position contingent on contract award. Must be able to obtain and maintain applicable Census Special Sworn Status and Personal Identity Verification (PIV) credentials, follow confidentiality requirements, and support Census core business hours and coordinated critical processing runs.
Benefits
Benefits
- 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
Skills Required
- Bachelor’s degree in data science, statistics, computer science, mathematics, or a related field, or equivalent relevant professional experience
- 8+ years of experience in applied data science or statistical analysis
- 3+ years of experience with NLP or text classification
- Hands-on experience applying large language models
- Strong Python and SQL skills
- Experience preparing and analyzing large datasets using statistical and machine learning libraries
- Strong experimental design and model evaluation skills, including sampling, class imbalance, error analysis, and baseline comparisons
- Experience supporting deployed models and collaborating with engineers
- Ability to explain analytical findings and limitations to technical and nontechnical audiences
- Advanced degree in a relevant field
- Experience with survey data, industry and occupation coding, or detailed classification taxonomies
- Experience with AWS-based analytical workflows
- Experience in federal statistical or sensitive-data environments
- Ability to obtain and maintain Census Special Sworn Status and Personal Identity Verification credentials
- Ability to follow confidentiality requirements and support Census core business hours and coordinated critical processing runs
What We Do
Passion and Purpose. We are a start-up borne out of a group of architects and engineers providing solutions that transform with digital breakthroughs. We Partner - Partnership to us means being all in - on the ground, in the weeds, producing and delivering mission solutions. We Innovate - driving digital innovation and experimentation to improve decision services for the American people. We Engage – co-creating to deliver HCD (human centered design) to solve tour customers digital challenges. We Deliver – and take pride and accountability in everything we do. We are Ignite IT. Ignite IT deploys teams that are customer-obsessed, delivering a CX (Customer Experience) renaissance to our Private and Public Sector customers. We combine Commercial BX (Business of Experience) with customer delivery experts to benefit agencies to innovate and improve services and launch new digital solutions. We have built a company foundation dependent on each of our employee’s success, and provide a healthy, engaging, community where learning and knowledge sharing are encouraged. We empower our employees to be bold, continually innovate and propose new ideas to improve processes and services. We engage daily on work that touches millions of people, always focusing on positive impacts and results. We enable mission services that enable Enterprise companies to succeed in digital transformation, protect Border Agents, deploy the Army Reserve and National Guard, providing decision dominance to make America safer. We apply these digital services across the Private and Public Sectors to include include HHS, DHS, AOUSC and Army. We aim to continually raise the bar for our customers to expect more from their digital services. For more information, visit www.igniteitservices.com
