Terms of Reference: AI Development and Integration Expert

/by

Background RNW Media 

RNW Media is an international digital media organisation working on social change. We believe in a world in which young people, in all their diversity, confidently claim their rights, assume their place in society and drive social change. Our mission is to accelerate the impact of changemakers by co-creating youth-centred digital media solutions built on data insights and mutual learning experiences. Owned by RNW Media, the “Love Matters Homebase” is a platform that a acts as a content bank which allows its members to contribute, share knowledge, co-create online SRHR/CSE content and simultaneously to ensure content is held to the highest quality standards, both in terms of CSE information & digital learning formats. 

Objective 

RNW Media is planning to pivot the Love Matters Homebase to become an AI-powered centralized platform for SRHR education and advocacy, powered by a distinctive embedded LLM. By incorporating advanced data aggregation technology, the platform will: 

  1. Aggregate comprehensive and up-to-date SRHR educational and advocacy materials from diverse global sources and provide ready to use output for SRHR.
  2. Ensure content respects cultural sensitivities and addresses local needs.
  3. Utilize machine learning to analyze usage patterns and improve content delivery.

Scope of the Assignment 

  • Establish the foundational infrastructure for data aggregation: Set up requirements and train a large language model that will provide accurate and current SRHR information for end users. 
  • Implement AI-driven content curation and customization features:  Create an API for the large language model   
  • Enhance content dissemination capabilities and improve user engagement: Personalize the content on the user feed via an automated tagging system and recommender algorithm integrated into the raw data pipeline. 
  • Documentation: Provide the needed technical documentation for integrating the API for possible further implementations. 

Key Deliverables 

  1. Integrated LLM into the Love Matters Homebase: A new user interface positioned on the Love Matters Homebase web application, providing interactive chatbot features in addition to features like personalized suggested searches and outbound links for referencing when providing answers. 
  1. New personalized feed: Renewed feed with an additional feed displaying personalized prompt recommendations with the help of a tagging and recommendation system. 

Activities 

  1. Data Aggregation Infrastructure Setup: 4 Weeks

   – Together with an SRHR expert, discuss and list resources that can be used as a source to train the model.  

   – Implement APIs for data retrieval from various global sources.  

   – Develop a robust data pipeline for extracting, transforming, and loading SRHR content from up-to-date open-source databases and repositories. 

   – Set up cloud infrastructure for scalable data storage and processing.  

  1. Development (Vendor): 2 Weeks

   –  Integrate machine learning models to categorize and tag aggregated content based on relevance (Testing and feedback will be provided by RNW Media staff).   

   – Develop algorithms to personalize content recommendations based on user behavior and preferences. 

  1. Launch and Post-Launch Activities (RNW Media & Vendor): 1 Week

   – Develop APIs for content delivery, enabling seamless integration with other platforms and social media channels.  

   – Implement analytics tools to track user engagement and content performance.  

    -Launch the features on a testing environment.  

  1. Testing and Quality Assurance (RNW Media): 1 Week

   – Perform functional testing to ensure all features work as intended.  

   – Conduct cross-browser and cross-device compatibility testing.  

   – Validate the new pages’ accessibility and compliance with web standards.  

   – Identify and fix bugs and issues.  

Timeline 

  • Start date: 25 November 2024 
  • Initial assessment deadline before final QA: 20 January 2025 
  • Final deadline: 27 January 2025 

Required Skills and Experience 

To deliver the aforementioned activities, we expect that the following skills and experience will be required.   

Large Language Model (LLM) Expertise: 

   – Proficiency in working with large language models (e.g., GPT, BERT) and experience in fine-tuning or adapting these models for specific use cases. 

   – Familiarity with multilingual model implementation, particularly for English and French. 

Data Engineering and Integration: 

   – Experience with data extraction, transformation, and loading (ETL) from both APIs and unstructured data sources, especially open-source SRHR-related content. 

   – Ability to develop robust, scalable data pipelines tailored for real-time data aggregation. 

API Development: 

   – Experience creating standalone APIs for LLM integration, ensuring secure handling, especially with SSL requirements in place. 

   – Knowledge of best practices for designing and documenting APIs for future scalability and integration. 

Personalization and Recommendation Systems: 

   – Proficiency in designing and implementing recommendation algorithms based on user behavior and automated tagging for personalized content feeds. 

   – Familiarity with machine learning libraries (e.g., TensorFlow, PyTorch, Scikit-Learn) and flexibility in framework choice based on project requirements. 

Compliance and Content Sensitivity: 

   – Familiarity with automated content moderation tools or techniques to ensure outputs respect cultural and regulatory sensitivities specific to SRHR content. 

Analytics and User Engagement Tracking: 

   – Experience setting up and managing analytics environments, ideally with proficiency in the Google Analytics suite for tracking user engagement and content performance. 

   – Ability to interpret data insights to iteratively improve content recommendation and user experience. 

Documentation and Communication: 

   – Strong skills in documenting technical processes, particularly for API integration and data aggregation pipelines. 

   – Excellent communication skills to work closely with the in-house product manager and SRHR experts for tailored testing and data gathering. 

Application 

To apply, send in: 

  • Your offer including a proposed timeline (if you fully agree with the above, you can use that one), budget, key contacts and short explanation of your background and services. 
  • Examples of past work (portfolio including similar web design and/or development projects). 
  • An additional online interview might be required if necessary. 
  • Interested applicants are invited to submit their proposals to noyan.er@rnw.org. 

In case you have any questions about this assignment do not hesitate to contact Noyan Er (Digital Product Manager).