Terms of Reference: GenAI Narrative, Content & Evaluation Specialist; Individual Consultant

Who We Are

Girl Effect is an international non-profit that builds media that girls want, trust and need.

From chatbots to chat shows and TV dramas to tech, our content helps adolescent girls in Africa and Asia make choices and changes in their lives, igniting their confidence to act differently at a time that can define their future.

We create safe spaces for girls, sharing facts, and answering questions about health, nutrition, education, relationships, and so much more. To arm girls with the skills to negotiate and redefine what they are told is possible "for a girl".

Our reach is 50 million and counting. And we're using technology to reach girls at scale so every girl can choose to be in control of her body, health, learning and livelihood.

Because when a girl unlocks her power to make different choices that change her life, it inspires others to do so too. She starts a ripple effect that impacts her family, community, and country.

That's the Girl Effect.

Over the past five years, Girl Effect has pioneered the development of digital products using conversational technologies to connect girls with critical health information. In particular, we have made significant strides in our South Africa-based chatbot, Big Sis, building new AI and GenAI capabilities, providing essential information on sensitive topics more seamlessly than ever before. We are now entering our most ambitious stage:
Mental Health Anytime Anywhere. This initiative moves beyond simple generative AI to predictive, agentic systems capable of anticipating a user's readiness to act and delivering personalized pathways for real behavior change.

This new phase involves two key components:

• Mental Health Agent: A system capable of conducting single-session therapy and supporting users in need, developed in collaboration with clinical psychologists.

• SBCC Agent: A Social and Behavior Change Communication agent that leverages reasoning Large Language Models (LLMs) to proactively guide users through behavioral obstacles. This agent will identify barriers and dynamically select strategies to "nudge" users toward positive health outcomes, such as service uptake.

The Consultant will bridge the gap between creative content, psychological safety, and technical implementation. You will ensure our AI agents speak with the right voice (especially for the South African context), adhere to safety protocols, and effectively guide users through complex health journeys.

• Data and Evaluation Curation:
  You will curate and refine datasets used to train and test our models. This includes generating and labeling sentences to ensure high-quality, culturally relevant inputs, and creating "gold standard" datasets for benchmarking performance.
• Prompt Engineering:
  You will design and optimize system prompts that control the persona, tone, and safety guardrails of the Mental Health and SBCC agents. This involves translating program needs into instructions that allow the LLM to act as an empathetic advisor and guide.
• Quality Assurance & System Analysis:
  Using tools like Lang Smith, you will conduct system analysis to trace model reasoning and identify failure modes (e.g., safety degradation). You will monitor "out of topic" queries to refine how the bot handles unexpected inputs.
• Testing, Evaluation, and Writing:
  You will evaluate chatbot responses for subjective parameters like tone, empathy, and cultural relevance, as well as objective accuracy. You will conduct rigorous testing to ensure the agents meet safety metrics and document these findings to inform technical iterations.

The Chatbot Content writer should have experience creating content for and interacting with youth audiences in South Africa. The Chatbot Content writer will use their familiarity with this demographic and the way they use language, including slang, to inform our AI and ML language models for the purposes of safeguarding, imparting knowledge, and speaking in a tone and language that resonates with young people.

They will also be central to creating and curating the content that shapes the large language models Girl Effect implements in its chatbots. This will include evaluating the responses of our generative AI chatbot and writing and curating prompts that produce tone and cultural relevance that our audience will resonate with.

• The consultant will be expected to be available for 1.5 days (12 hours) per week through the time period of February 2026 to June 2026.

The consultant will be remote and must be conducive to working with teams in the UK, USA, Sub-Saharan Africa & India synchronously and asynchronously.

• Terms of reference published:
  February 5, 2026
• Deadline for responses:
  February 18, 2026
• Project commencement:
  March 2026

Specific tasks and deliverables may vary, but example deliverables may include:

• Evaluation Reports:
  Detailed reports analyzing model performance against safety and behavioral goals.
• Prompts for…