In A Nutshell
Wadhwani AI is a non-profit institute building and deploying applied AI solutions to solve critical issues in public health, agriculture, education, and urban development in underserved communities in the global south. We collaborate with governments, social sector organizations, academic and research institutions, and domain experts to identify real-world problems, and develop practical AI solutions to tackle these issues with the aim of making a substantial positive impact.
Responsibilities
- Annotate user queries sequentially within each session while considering previous queries and responses to maintain conversational continuity and context.
- Translate corrected user queries into clear and grammatically accurate English while preserving the original meaning and intent.
- Rephrase user queries into complete standalone English queries using the current query, chat history, and available metadata for effective answer generation.
- Identify and map crops mentioned in queries to the standardized crop list and assign the appropriate advisory category based on query intent and context.
- Ensure consistency, accuracy, completeness, and adherence to project annotation guidelines, SOPs, confidentiality, and quality standards.
- Review annotated sessions to validate translations, rephrased queries, crop mapping, advisory category mapping, and contextual understanding incorporated during annotation.
- Identify inconsistencies, ambiguities, missing information, annotation errors, and potential biases, and provide corrective feedback to improve annotation quality and dataset reliability.
- Ensure standardized interpretation and uniformity across the dataset, with each session handled by a single annotator and independently reviewed for quality assurance.
- Validate agricultural advisories, pest management protocols, and crop-related recommendations against established agronomic literature and extension advisories.
- Create and maintain high-quality ground-truth datasets, document evaluation decisions, and follow standardized evaluation protocols.
- Evaluate AI-generated responses for accuracy, relevance, completeness, and alignment with recommended agricultural practices.
- Identify incorrect, incomplete, or misleading AI responses, flag knowledge gaps, and provide structured feedback to support model refinement and improvement.
- Participate in calibration and review discussions to maintain evaluation consistency and support continuous enhancement of the AI system.
- Perform additional tasks related to annotation, review, evaluation, and quality assurance as assigned by the supervisor or project team.
Skillset
- Education: Master’s degree (Second or Third or Fourth year) or PhD (Any year) in Entomology, Plant Pathology or Agronomy from an accredited university or college. Students pursuing a degree can also participate in.
- Experience: Experience working with or training farmers is required.
- Domain Knowledge: Strong understanding of crops, common plant diseases, pest identification, and agronomic terminology. Ability to assess the correctness of crop-related annotations is essential.
- Tools Availability: Access to a functional laptop and a reliable internet connection is mandatory.
- Technical Skills: Proficient with multitasking in a split-screen setup, using Excel or Google Sheets, and performing data entry and copy-paste operations efficiently.
- Communication Skills: Strong verbal and written communication skills to participate in team calls, deliver feedback effectively, and coordinate with the project team.
- Collaboration: Ability to work effectively with geographically dispersed teams and in multicultural environments.
