Code for Africa (CfA) has announced applications for the Lexicon Fellowship, inviting South Africa-based computational linguists and natural language processing (NLP) practitioners to contribute innovative research tackling hate speech and harmful online discourse. The three-month part-time fellowship aims to strengthen information integrity by developing a structured hate speech dictionary to support AI-powered human rights monitoring.
The application acceptance period is as follows July 13, 2026successful fellows will receive a monthly stipend, mentorship, and the opportunity to work with leading researchers and human rights organizations.
Fellowship supports AI solutions for human rights monitoring
The Lexicon Fellowship is part of a joint initiative between Code for Africa and the Norwegian Ministry of Foreign Affairs to improve the detection and monitoring of online hate speech across South Africa.
The program is seeking the following experts who can identify and analyze weaponized language:
- hate speech.
- Incitement to violence.
- Harmful online narratives.
- Defamatory or discriminatory language.
- Coded expressions and new hate terms.
- Slogans and hashtags linked to coordinated online campaigns.
This initiative focuses on several priority themes including:
- City council elections.
- xenophobia.
- LGBTQ+ community.
- Gender-based hatred and online violence.
- Protection of ethnic minorities.
By creating structured, machine-readable dictionaries, fellows contribute tools that help police online spaces while respecting international human rights principles.
Building NLP models for low-resource African languages
The main objective of this fellowship is to strengthen natural language processing capabilities for African languages, especially those with limited digital language resources.
Selected fellows will work directly with:
- Tokenization technology.
- Word embedding.
- Classification pipeline.
- Machine learning workflow.
- Transformer-based language model.
- Low-resource language processing.
The fellowship also encourages participants to develop models that can identify new hate speech that may not appear in existing datasets. Rather than relying solely on predefined vocabularies, fellows will design flexible classification systems that continually detect new slurs, coded language, and evolving harmful narratives.
Responsibilities of selected fellows
Participants will play a key role in developing a high-quality hate speech dictionary for research and monitoring purposes.
Key responsibilities include:
- Collect and verify words, phrases, slogans, slang, and coded language related to hate speech.
- Classify harmful speech using an internationally recognized human rights framework.
- Ensures datasets remain machine-readable and suitable for AI applications.
- Supports the development of NLP models to detect harmful online content.
- Contributes to analysis tools such as:
- Polarization score.
- Disinformation Vulnerability Index.
- Early warning indicators of online violence.
The fellowship has built-in ethical safeguards to ensure that cultural context, bias, and freedom of expression are taken into account in language analysis, while adhering to international standards such as the Rabat Plan of Action and the International Covenant on Civil and Political Rights (ICCPR).
Diverse data sources inform your research
To create a comprehensive hate speech dictionary, fellows analyze information collected from multiple online and offline sources.
These include:
- Facebook.
- TikTok.
- X.
- YouTube.
- telegram.
- Whatsapp.
- Archives of historical events.
- Human Rights Report.
- This is an election observation report.
- Fact-checked news articles.
- Community reporting channel.
- Public speech.
- Focus group discussion.
- Key informant interviews.
This broad evidence base will enable the development of contextual AI systems that can recognize harmful narratives before they escalate into systemic violence and discrimination.
Fellowship benefits
Successful applicants will receive a variety of professional and financial benefits during the fellowship period.
These include:
- Competitive monthly salary.
- 3 month part time fellowship.
- Technical guidance from CfA’s iLab and TechLab teams.
- Access to regional innovation and research networks, including the Omdena regional branch.
- Cooperation with human rights organizations.
- Opportunity to contribute to real-world information integrity efforts across Africa.
The program also provides technical support for testing, refining, deploying, and expanding NLP models and hate speech detection systems.
Eligibility requirements
Code for Africa is seeking highly skilled NLP practitioners with strong technical and research backgrounds.
Applicants must have:
- Advanced Python programming skills.
- Extensive knowledge of Google Workspace.
- At least 2 years of experience in data science, preferably with a specialization in NLP.
- Experience with machine learning and NLP pipelines.
- Familiarity with transformer models and the Hugging Face ecosystem.
- Experience fine-tuning large language models using PEFT, LoRA, or QLoRA.
- Knowledge of encoder models such as Afro-XLMR and AfriBERTa.
- Experience adapting tokenization systems to morphologically rich African languages.
- Knowledge of collecting and analyzing social media datasets.
Additional benefits include experience working with low-resource South African languages and knowledge of regional politics, culture and conflict dynamics.
Applicants must also prove that:
- Be fluent in English.
- Be proficient in at least one South African language, including Afrikaans, IsiZulu, IsiXhosa, Sepedi, Sesotho, or Setswana.
- A commitment to advancing African language technology.
- A portfolio showcasing previous NLP or machine learning projects.
- Excellent communication and collaboration skills.
Application deadline
Interested candidates should submit their applications by the following deadlines: July 13, 2026.
Applicants selected for interviews will be contacted directly by Code for Africa following the screening process.
Advancing African language technology through citizen innovation
As Africa’s largest civic technology and open data incubator, Code for Africa continues to invest in innovative technologies that strengthen digital democracy, combat misinformation, and improve public accountability.
Through the Lexicon Fellowship, the organization aims to support human rights monitoring, protect vulnerable communities, and strengthen information resilience across South Africa, while advancing AI applications in African languages.
For more opportunities like this, please visit us on Facebook, Instagram, TwitterLinkedIn and WPChannel
Disclaimer: Global South Opportunities (GSO) is not the organization providing this opportunity. For inquiries, please contact the official organization directly. Please do not submit your application or resume to GSO as it cannot be processed by GSO. Due to the large volume of emails we receive every day, we may not be able to respond to all inquiries. Thank you for your understanding.
