We are seeking aPrompt Engineer design, test, and refine interaction patterns for generative AI systems used across our client engagements. This role is central to shaping how users experience AI-powered applications and ensuring that LLM outputs are relevant, reliable, safe, and aligned to mission needs. The Prompt Engineer will work closely with AI Developers,LLMOpsEngineers, Evaluation Scientists, and product/design teams to craft prompts, structure context, improve retrieval strategies, and create intuitive multi-turn interaction flows.

• Design, iterate, andoptimizeprompts, templates, and prompt chains that improve the accuracy, reasoning, and usability of generative AI outputs.
• Develop context management strategies, including document selection, summarization, formatting, role instructions, andretrievalaugmentation patterns.
• Partner with developers to implement prompt engineering techniques within RAG pipelines, agentic workflows, evaluation harnesses, andapplicationUX layers.
• Conduct structured experimentsincluding A/B tests and rubric-based evaluationsto measure prompt performance andidentifyopportunities for improvement.
• Collaborate with AI Evaluation Scientists to defineevaluationmetrics tied to quality, relevance, consistency, safety, and hallucination reduction.
• Work closely with UX and product design teams to develop interaction flows that guide user inputs, clarify system capabilities, and reduce cognitive load.
• Help translate complex subject-matter content into structured, LLM-friendly formats that improve system reasoning and reduce error rates.
• Support the design of safe output behaviors by implementing guardrails, filtering strategies, and context constraints aligned with governance principles.
• Document prompt patterns, style guides, failure modes, and best practices for internal reuse and team enablement.
• Stay up to date on emerging prompting strategies (e.g.,CoT, self-critique, toolusescaffolding, function calling, agentic behaviors) and share insights with the team.
• You will contribute to the growth of our AI & Data Exploitation Practice!

• Ability to hold a position of public trust with the U.S. government.
• Bachelor27s degree in Computer Science, Linguistics, Cognitive Science, Data Science, ora relateddiscipline.
• 5+ years of total experience.
• 2+ years of experience in AI/ML, NLP, content design, UX writing, or similar roles involving structured interaction design.
• Demonstrated experience designing and testingprompts forLLMs in real applications (GPT-style models, Claude, Gemini, Llama, etc.).
• Familiarity with RAG architectures, vector databases, embeddings, and context retrieval strategies.
• Strong analytical skills with the ability to run structured experiments, analyze model outputs, and document findings.
• Proficiencywith Python and common AI frameworks (e.g.,LangChain,LlamaIndex) is preferred.
• Experience with UX design principles, information architecture, or conversational design is highly valuable.
• Understanding ofresponsible AI concepts, including safety, privacy, bias, and governance-aligned prompting strategies.
• Excellent written and verbal communication skills with the ability to translate domain-specific content into clear, effective AI instructions.
• Ability to collaborate across engineering, product, governance, and evaluation teams to shape coherent user-AI interactions.

Steampunk relies on several factors to determine salary, including but not limited to geographic location, contractual requirements, education, knowledge, skills, competencies, and experience. The projected compensation range for this position is $115,000 to $140,000. The estimate displayed represents a typical annual salary range for this position. Annual salary is just one aspect of Steampunk27s total compensation package for employees. Learn more about additional Steampunk benefits here.

Identity Statement

As part of the application process, you are expected to be on camera during interviews and assessments. We reserve the right to take your picture to verify your identity and prevent fraud.