AI Prompt Engineer — GenAI Solutions for Government

CACI International; Washington (District of Columbia); Sur place; USD 53 000 - 106 000; Plein temps

Résumé du poste

CACI International is seeking an AI Prompt Engineer to join the BEAGLE ASF team supporting CBP in the DMV area. You will design, test, and optimize prompts for large language models to deliver high-quality outputs across government-use cases.

Ideal candidates have hands-on experience with LLMs, Python or JavaScript/TypeScript, and familiarity with few-shot and chain-of-thought prompting, as part of an Agile delivery approach.

Qualifications

• U.S. citizenship; ability to pass CBP background check.
• 3-year felony check; 1-year illegal drug check; 1-year misconduct check.
• Onsite in Ashburn, VA or Washington, DC.
• Bachelor’s degree in CS/SE/IT or related field (or equivalent).
• 0–3 years software development experience.
• 1+ year hands-on with LLMs or generative AI.
• Experience with Python or JavaScript/TypeScript.
• Familiar with few-shot, chain-of-thought, structured outputs, function calling.

Responsabilités

• Prompt design, optimization, and refinement for LLMs.
• Iterative testing and result analysis; drive improvements.
• Collaborate with groups to integrate prompts into workflows.
• Research latest LLM and prompt engineering techniques.
• Support fine-tuning and Retrieval Augmented Generation (RAG).
• Debug unexpected model behaviors and root causes.

Connaissances

LLM experience

Programming: Python/JavaScript/TypeSct

Prompt engineering

Agile/Scrum

Formation

Bachelor's degree in Computer Science / Software Engineering / Information Technology

Outils

OpenAI GPT series

Llama

Gemini

Description du poste

CACI International is seeking an AI Prompt Engineer to join the BEAGLE ASF team supporting CBP in the DMV area. You will design, test, and optimize prompts for large language models to deliver high-quality outputs across government-use cases.

Ideal candidates have hands-on experience with LLMs, Python or JavaScript/TypeScript, and familiarity with few-shot and chain-of-thought prompting, as part of an Agile delivery approach.