Chris Denq (they/he)

I am a full-time data scientist and part-time academic researcher committed to social impact and safety.

Data Science
Operationally, my work tackles administrative challenges that lie at the intersection of safety and public policy. Broadly, I work with predictive modeling and simulations to inform policy makers, research managers, and executive stakeholders at the federal level. Prior to my 2 years in government, I worked for 2 years in the tech and financial sectors in various data roles.

Research
My research interests lie at the intersection of computer science and law/policy.

• On the computer science front, I am interested in improving natural language processing (NLP) capabilities of large language models (LLMs), with the overall aim of refining machine reasoning for downstream legal applications. Specifically, I am interested in understanding how we can better ground LLMs to sources of truth, and am currently investigating this architecturally via retrieval-augmented generation (RAG) and mixture-of-expert (MoE) systems.
• On the law/policy side, I am doubly interested in both the (1) ethical regulation and (2) socially impactful deployment of LLMs, particularly in underserved populations.
  • For regulation: I am interested in updating existing legal doctrine to account for novel LLM misuse (e.g. legal disincentives for new types of LLM attacks).
  • For deployment: I am interested in leveraging LLMs in accelerating mundane federal cases (e.g. mass adjudication), particularly towards a more efficient allocation of civil benefits.