HausaNLP

HausaNLP

Timeline

Timeline

Sept. 2024 - Dec. 2024

Sept. 2024 - Dec. 2024

Role

Role

Product Designer

Product Designer

Tools

Tools

Figma

Figma

Team

Team

PM: Vy | Tech Lead: Benji | Developers: Bhuvana, Evan, Siri, Keshav, Sophie

PM: Vy | Tech Lead: Benji | Developers: Bhuvana, Evan, Siri, Keshav, Sophie

Context

Context

Overview

Overview

I designed a platform for HausaNLP, an open-source project advancing NLP for Hausa — one of Africa’s most spoken languages. Starting with just a logo, I led the end-to-end design to help researchers access data, contribute resources, and increase Hausa representation in AI.

I designed a platform for HausaNLP, an open-source project advancing NLP for Hausa — one of Africa’s most spoken languages. Starting with just a logo, I led the end-to-end design to help researchers access data, contribute resources, and increase Hausa representation in AI.

Problem & Solution

Problem & Solution

Why Hausa Deserves Better Tools

Why Hausa Deserves Better Tools

Problem: Currently, Hausa lacks critical natural language processing resources, limiting its representation in AI technologies and reinforcing Western biases in existing models.

Solution: A platform for HausaNLP to provide open-source models and datasets, ensuring Hausa’s inclusion in emerging AI technologies and promoting linguistic diversity in NLP.

Planning

Moving Fast Without Losing Purpose

Moving Fast Without Losing Purpose

I focused on designing features that addressed real challenges in Hausa NLP, making the platform useful for researchers from day one. With limited time and many core features to build, we prioritized speed and user needs over extended ideation.

I focused on designing features that addressed real challenges in Hausa NLP, making the platform useful for researchers from day one. With limited time and many core features to build, we prioritized speed and user needs over extended ideation.

Understanding the users

Understanding the users

Who are we creating the product for?

User Persona 1 - Researcher

Dr. Amina Yusuf

Computational Linguist & NLP Researcher

Pain Points

Pain Points

  • Limited availability of high-quality Hausa datasets

  • Difficulty locating niche or specific data types

  • Unclear submission requirements when contributing datasets

  • Limited availability of high-quality Hausa datasets

  • Difficulty locating niche or specific data types

  • Unclear submission requirements when contributing datasets

Needs

Needs

Amina wants to easily find and filter Hausa NLP datasets for her research, contribute her own data with a simple upload process, and receive clear feedback on submissions. She values quick access to relevant resources and staying informed about HausaNLP’s mission.

Amina wants to easily find and filter Hausa NLP datasets for her research, contribute her own data with a simple upload process, and receive clear feedback on submissions. She values quick access to relevant resources and staying informed about HausaNLP’s mission.

User Persona 2 - Administrator

Ibrahim Bello

HausaNLP Data Curator & Admin

Pain Points

Pain Points

  • Managing high volume of submissions with limited automation

  • Ensuring data consistency and accuracy

  • Limited tools for effective communication with contributors

  • Managing high volume of submissions with limited automation

  • Ensuring data consistency and accuracy

  • Limited tools for effective communication with contributors

Needs

Needs

Ibrahim’s focus is on efficiently reviewing dataset submissions, maintaining data quality, and managing the catalog. He needs streamlined tools for approving, denying, and editing entries, along with clear workflows for providing feedback to contributors.

Ibrahim’s focus is on efficiently reviewing dataset submissions, maintaining data quality, and managing the catalog. He needs streamlined tools for approving, denying, and editing entries, along with clear workflows for providing feedback to contributors.

Branding

Building a Scalable Design System for Low-Resource Tools

For this project all we were given was the logo, I based our color palette on that, mostly a mix of blue with pops of colors to convey certain emotions. For the font we aimed for something legible and we choose poppins a sans serif font, that is slightly rounded, yet minimal.

Considerations

Going beyond pretty interfaces

Going beyond pretty interfaces

  1. Starting with minimal resources, I broke the challenge into smaller goals: simplifying NLP tools, ensuring accessibility, and addressing user engagement.

  1. Recognizing a diverse audience of researchers and developers, I designed with clarity and intuitiveness in mind to support varying technical expertise.

  1. NLP platforms often present complex technical data, so I focused on a clean layout, clear labels, and call-to-actions to reduce cognitive load.

  1. Without direct Hausa-language competitors, I adapted features from existing platforms but tailored them to fit the unique needs of HausaNLP.

High-Fidelity Wireframes

High-Fidelity Wireframes

A Platform that Speaks the Community’s Language

A Platform that Speaks the Community’s Language

Researchers (P0)

Data Catalog

Data Catalog

  1. Clean Table Layout: A polished, browsable table format enables quick dataset scanning, improving usability and efficiency.


  1. Advanced Filtering: Filters let users narrow datasets by specific criteria, saving time and streamlining research.


  1. Search Bar: Instant search functionality supports both exploratory and targeted dataset discovery.

  1. Clean Table Layout: A polished, browsable table format enables quick dataset scanning, improving usability and efficiency.


  1. Advanced Filtering: Filters let users narrow datasets by specific criteria, saving time and streamlining research.


  1. Search Bar: Instant search functionality supports both exploratory and targeted dataset discovery.

Upload Form

Upload Form

  1. Stepper: A multi-step layout simplifies the upload process, reducing overwhelm and ensuring accurate submissions into manageable sections.


  1. Focused Questioning: Curated fields capture essential details while minimizing unnecessary inputs.


  1. User Guidance: Tooltips and clear instructions make the form accessible for all users.

  1. Stepper: A multi-step layout simplifies the upload process, reducing overwhelm and ensuring accurate submissions into manageable sections.


  1. Focused Questioning: Curated fields capture essential details while minimizing unnecessary inputs.


  1. User Guidance: Tooltips and clear instructions make the form accessible for all users.

Researchers (P1)

Statistics

Statistics

  1. Data Visualization: Clear charts and graphs present complex metrics in an easily understandable format.


  1. Interactive Features: Filters and hover details let users explore specific data points for deeper insights.


  1. Customizable Filters: Users can filter data by year, language, and tasks, enabling tailored analysis

  1. Data Visualization: Clear charts and graphs present complex metrics in an easily understandable format.


  1. Interactive Features: Filters and hover details let users explore specific data points for deeper insights.


  1. Customizable Filters: Users can filter data by year, language, and tasks, enabling tailored analysis

Contributors

Contributors

  1. Researcher Spotlight: Highlights individual contributors and teams, celebrating their work and fostering community.


  1. Detailed Profiles: Displays affiliations, contributions, and achievements, adding credibility and context.

  1. Researcher Spotlight: Highlights individual contributors and teams, celebrating their work and fostering community.


  1. Detailed Profiles: Displays affiliations, contributions, and achievements, adding credibility and context.

FAQs

FAQs

  1. Clean Table Layout: A polished, browsable table format enables quick dataset scanning, improving usability and efficiency.


  1. Advanced Filtering: Filters let users narrow datasets by specific criteria, saving time and streamlining research.


  1. Search Bar: Instant search functionality supports both exploratory and targeted dataset discovery.

  1. Clean Table Layout: A polished, browsable table format enables quick dataset scanning, improving usability and efficiency.


  1. Advanced Filtering: Filters let users narrow datasets by specific criteria, saving time and streamlining research.


  1. Search Bar: Instant search functionality supports both exploratory and targeted dataset discovery.

Admin (P0)

Data Catalog

Data Catalog

  1. Edit and Delete Functionality: Admins can update or remove datasets to maintain accuracy and relevance.


  1. Edit Modal: A dedicated modal ensures controlled and focused editing, reducing errors and improving usability.


  1. Intentional Actions: The "Clear Form" requirement prevents accidental changes, adding a layer of security.

  1. Edit and Delete Functionality: Admins can update or remove datasets to maintain accuracy and relevance.


  1. Edit Modal: A dedicated modal ensures controlled and focused editing, reducing errors and improving usability.


  1. Intentional Actions: The "Clear Form" requirement prevents accidental changes, adding a layer of security.

Contributors

Contributors

  1. Researcher Spotlight: Highlights individual contributors and teams, celebrating their work and fostering community.


  1. Detailed Profiles: Displays affiliations, contributions, and achievements, adding credibility and context.

  1. Researcher Spotlight: Highlights individual contributors and teams, celebrating their work and fostering community.


  1. Detailed Profiles: Displays affiliations, contributions, and achievements, adding credibility and context.

Admin (P1)

Contributors

Contributors

  1. Transparent Communication: Admins can send detailed emails to users explaining why their submission was denied.


  1. Constructive Feedback: Provides users with actionable insights to improve future submissions.


  1. Streamlined Process: A clear, intuitive interface makes composing and sending emails quick and efficient.

  1. Transparent Communication: Admins can send detailed emails to users explaining why their submission was denied.


  1. Constructive Feedback: Provides users with actionable insights to improve future submissions.


  1. Streamlined Process: A clear, intuitive interface makes composing and sending emails quick and efficient.

Conclusion

Conclusion

Reflections & Impact

Reflections & Impact

This project challenged me to design independently for a global client and adapt to diverse user needs, from cultural context to varying literacy levels. It was especially meaningful knowing the work could impact an underrepresented community.


With more time, I would have conducted user testing to better understand end-user needs beyond client feedback.

Azaan Shaikh

👋 Thank you for stopping by :)

Azaan Shaikh

👋 Thank you for stopping by :)