Jobright is your personal AI job search agent transforming the job search process. They are seeking an early-career Prompt Engineer to craft and evaluate prompts that enhance the performance of AI outputs across user-facing features.

Why Join Us

• Build real, production AI agents used by real users
• High ownership and impact
• Work at the intersection of AI, agents, and product
• Shape how people experience AI-driven job search

Responsibilities

• Craft, test, and iterate on prompts and prompt chains that drive the quality and reliability of AI agent outputs across user-facing features
• Develop systematic evaluation frameworks to measure prompt performance against metrics like accuracy, consistency, tone, and hallucination rate
• Investigate failure modes and edge cases in model responses, diagnose root causes, and design prompt-level fixes that scale across use cases
• Maintain a versioned prompt library with clear documentation, so the team can track what changed, why, and how it performed
• Work closely with product and engineering teams to translate ambiguous user needs into structured instructions that LLMs can follow reliably

Qualification

Required

• Recent graduate or early-career professional (0–2 years of experience) with a degree in Computer Science, Linguistics, Cognitive Science, or a related field
• Deep curiosity about how LLMs interpret and respond to instructions, and an intuitive sense for why small wording changes can produce dramatically different outputs
• Proficiency in Python for scripting evaluation pipelines, parsing model outputs, and running batch prompt experiments at scale
• Exceptional written communication skills — the ability to write instructions that are precise, unambiguous, and robust to misinterpretation

Preferred

• Previous experience designing prompts for production applications, not just casual ChatGPT usage — understanding the difference between a prompt that works once and one that works at scale
• Familiarity with techniques like few-shot prompting, chain-of-thought, self-consistency, and retrieval-augmented generation
• Hands-on exposure to evaluation tooling or building custom scoring rubrics to grade model outputs beyond simple pass/fail
• A genuine obsession with language — the kind of person who notices when a comma changes the meaning of a sentence