Senior AI/ML Engineer

Ignite•Suitland, MD

About The Position

Ignite IT Hub is seeking a senior AI/ML Engineer to maintain and enhance the U.S. Census Bureau's LLM Autocoder, which assigns standardized codes to survey write-in responses. You will build and test model and software improvements for the American Community Survey (ACS), develop pilots for additional surveys and fields, and work with data scientists and MLOps engineers to integrate approved changes into production.

Requirements

Bachelor's degree in computer science, engineering, data science, or a related field, or equivalent relevant professional experience.
8+ years in software or ML engineering, including 3+ years building NLP or ML applications and hands-on experience integrating LLMs into working systems.
Strong Python development skills and experience with data processing, APIs, automated testing, Git, and maintainable software design.
Experience with text classification, prompt design, structured model outputs, and measuring the effects of implementation changes on model performance.
Experience deploying or supporting applications in AWS, troubleshooting production issues, and collaborating effectively with data scientists and operational teams.

Nice To Haves

Experience with fuzzy matching, dictionary-based coding, survey text, or industry and occupation classification.
Experience with ML frameworks such as scikit-learn, PyTorch, or Hugging Face, and handling sensitive data in controlled federal environments.

Responsibilities

Develop and maintain Python components for text preprocessing, model invocation, prompt management, structured output validation, and coding workflows.
Implement approved changes to prompts, model configuration, dictionaries, rules, and supporting code in small, version-controlled increments.
Build pilots using LLMs, fuzzy matching, direct-match dictionaries, or hybrid approaches; adapt implementations to survey-specific taxonomies and requirements.
Write unit, integration, regression, negative, and edge-case tests, and work with the evaluation team to compare candidate results with approved baselines.
Reproduce production problems, isolate causes across code, data, prompts, and model behavior, and implement verified corrections.
Partner with the MLOps Engineer on packaging, deployment, rollback, and post-release verification within Government-approved AWS environments.
Participate in code reviews and SME working sessions; maintain technical documentation, user guidance, and reproducible examples for Government handoff.