Prompt Engineer (LLM Automation for Data Labeling & Localization)

About the Role

Innodata is building a team of prompt engineers to harness the power of LLMs to automate data annotation and human evaluation workflows. The goal is to facilitate accurate, localized, and culturally adapted data labeling and translation processes through effective prompt design and implementation. This team will collaborate directly with our client partner, a leading technology company, to identify opportunities for automation, design solutions, and drive measurable improvements. As a technical subject matter expert, you will work backwards from the customer problem statement to develop an efficient plan for execution.

You will collaborate with crossfunctional teams, including product managers, data scientists, and client teams, to solve complex problems, reduce human effort, and ensure that AIdriven processes meet high standards for quality and reliability. Your work will directly contribute to improving our clients data annotation and evaluation processes, enabling them to scale more efficiently.

Key Responsibilities

• Collaborate with data scientists, linguists, and localization experts to ensure accuracy and cultural relevance.

• • Prototype and validate AI models to demonstrate initial feasibility, potential impact, and overall effectiveness.

  • • Design, develop, and implement prompts for data labeling and localization processes within software applications.

    • • Understand the current components of the software stack, use cases and problems and iterate on solutions leveraging a solid knowledge of data structures, data formats, and data modeling.

      • • Conduct user testing and feedback analysis to optimize prompt design for data accuracy and linguistic consistency.

        • • Analyze model performance using key performance indicators (KPIs) and metrics, ensuring that AI models meet customer acceptance criteria and deliver highquality outputs.

          • • Communicate technical findings and solution strategies to both technical and nontechnical stakeholders, including presenting model performance and actionable insights in a clear, accessible manner.

            • • Collaborate on data pipelines and workflows that integrate LLMs into automated systems, enhancing both the efficiency and effectiveness of data annotation tasks.

              • • Create guidelines and training materials for prompt usage in data labeling and localization projects.

                • • Stay informed on data labeling and localization industry trends and tools to enhance prompt engineering techniques.