Code for Africa (CfA) is offering two fellowships aimed at combating the spread of hate speech through linguistic expertise and natural language processing (NLP). The fellowship will support the mapping and analysis of weaponised words, phrases, slogans, slang, and slurs associated with hate speech, harmful online narratives, and online violence against women, youth, and minority groups.
We are seeking experts based in Ethiopia who are interested in developing structured hate speech and incitement lexicons for human rights monitoring. The primary thematic focus areas include ethnicity, LGBTQ+ communities, women and youth and gendered hate.
The two-month, part-time fellowships are part of a joint initiative between CfA and the UN Office of the Special Envoy for the Horn of Africa (OSE-Horn) to strengthen information integrity by mapping indigenous weaponised words, phrases, slogans, slang, and slurs associated with hate speech and harmful online narratives.
What we are looking for is more than just a simple keyword list. A watchdog lexicon classification that is structured, version-controlled, and auditable language resource. It must systematically link terms, their variants, and contextual cues to defined harm categories, identified target groups, and assessed levels of incitement risk. In addition, it should be supported by clear annotation guidelines to ensure consistency, transparency, and accountability in how entries are classified and applied.
The selected fellows will be responsible for collecting, validating, and categorising words, phrases, slogans, slang, and slurs linked to hate speech and harmful narratives. They will ensure that the lexicons are machine-readable and aligned with international human rights standards, including the Rabat Plan of Action and the ICCPR proportionality principles.
The process will incorporate ethical safeguards to account for bias, cultural nuance, and context, while ensuring that the collection and classification of terms do not infringe upon freedom of speech and freedom of expression.
The mapping of hate lexicons is designed to support:
- Early warning systems for incitement to violence
- Detection of hate speech and discriminatory harassment
- Tracking patterns of dehumanisation, exclusion, ethnic cleansing rhetoric, and targeted attacks
- Identification of coordinated campaigns, including hashtags and slogan-based mobilisation
To guide the technical collection process, fellows will draw from multiple sources, including: Open social media platforms such as Facebook, TikTok, X, and YouTube; dark social platforms such as Telegram and WhatsApp; historical incident archives; human rights and election observer reports; reports on hate speech, ethnic violence, xenophobia, and extremism; fact-checked articles and web archives; tip lines and community reporting channels; rally speeches; focus Group Discussions (FGDs) and Key Informant Interviews (KIIs) and offline public forums.
Following the collection, validation, and categorisation process, the fellows will support the development of Natural Language Processing (NLP) models to detect harmful content. These models will leverage the structured lexicon classifications to generate analytical indicators such as a polarisation score and a disinformation vulnerability index.
The scoring will be informed by contextual factors, including how frequently and in what manner specific terms are used across social media platforms, whether they target particular groups, and whether their usage intensifies during high-risk periods such as elections, conflict, or protests.
Selected fellows will receive technical and research support from CfA’s forensic research team (iLab) and TechLab. This support will cover the resources required for testing, refining, deploying, and scaling the hate speech lexicons effectively.
Fellowship Package
- Stipend: Fellows will receive a competitive monthly stipend.
- Duration: Two months
- Mentorship: Fellows will receive guidance from CfA’s iLab and TechLab teams.
- Network: Opportunities to collaborate with regional innovation and research communities, including networks such as Omdena local chapters.
- Real-world impact: Fellows will work alongside human rights defender (HRD) organisations to implement practical solutions aimed at defending information integrity and strengthening information resilience.
Are You Eligible?
We are seeking professionals with:
- Technical skills: Advanced proficiency in Python and strong working knowledge of Google Workspace (G-Suite).
- NLP & Machine Learning: Solid understanding of NLP and machine learning pipelines, with hands-on experience in data science projects.
- Experience: A minimum of two years’ experience in data science, with a strong preference for candidates specialising in Natural Language Processing (NLP).
- Social media data collection: Familiarity with collecting and analysing data from platforms such as TikTok, Facebook, X, YouTube, and Telegram.
- Language proficiency: Fluency in Amharic and English, + ≥1 of the following languages: Afan-Oromo, Oromo, Somali and Tigrinya.
- Contextual understanding: Basic knowledge of cultural, political, and conflict dynamics within the specific regions.
- Portfolio evidence: Demonstrated experience through previous projects involving NLP and machine learning.
- Local language commitment: A clear commitment to advancing African local languages in technology and research.
- Conduct: Ability to maintain timely and open communication with the programme management team.
The deadline for applications is 17 April 2026.
To apply, complete this form.
Questions? Reach out to john.n@codeforafrica.org or jacktone.momanyi@codeforafrica.org.
About the partner:
Code for Africa is Africa’s largest civic technology and open data incubator. The organisation supports digital democracy and investigative data journalism initiatives to give citizens access to actionable information and amplify citizen voices to make society more responsive and accountable. CfA’s mission is to leverage technology and data to foster digital democracies and empower citizens by providing actionable information.
To all recruitment agencies: CfA does not accept agency resumes. Please do not forward resumes to our employment application line, CfA employees or any other CfA contact. CfA is not responsible for any fees related to unsolicited resumes.
Please note: Due to high volumes of applications, we are unable to respond to each one individually. If you are selected for an interview, we will contact you.
—————————————————————————————————————————