3 min read

Traversaal.ai Launches L.A.R.G.E: Pioneering AI for Underrepresented Languages

Traversaal.ai Launches L.A.R.G.E: Pioneering AI for Underrepresented Languages
Photo by Susan Wilkinson / Unsplash

At Traversaal.ai, we believe in a world where Artificial Intelligence speaks every language, understanding local nuances and cultural contexts. That's why we're thrilled to announce the launch of LARGE (Language AI Research for Global Equity), a dedicated R&D vertical within Traversaal focused on developing high-performance, state-of-the-art Large Language Models (LLMs) for non-English speaking regions. Our mission? To empower underrepresented languages and communities through cutting-edge AI that feels as natural as a conversation with a friend.

Building on Success: From Urdu Llama to New Horizons

Our journey began with Meta's APAC AI Accelerator program, where we earned recognition for building Urdu Llama, a state-of-the-art LLM specifically designed for the Urdu-speaking community. This model set a new standard, proving that AI could not only understand but also respect the subtleties and depth of regional languages. This success has inspired us to take our mission further, expanding our focus to new regions and languages that have historically been underserved by the AI industry.

Our next big milestone? Arabic Llama—a language model that will serve one of the world's most widely spoken languages, built from the ground up to reflect the cultural heritage, linguistic diversity, and unique nuances of Arabic speakers. With this and future models, we aim to deliver AI that is not only powerful but also deeply rooted in the authentic essence of each language and culture it represents.

The Team Behind LARGE

LARGE brings together a diverse and passionate team of researchers, engineers, and advisors, each dedicated to advancing the inclusivity and accessibility of AI. Our team is a blend of industry experts and academic researchers, each bringing unique insights to the table to ensure our models achieve both technical excellence and cultural relevance.

Advisors:

Core Team:

  • Ali Shafique: Head of engineering, leading model design and architecture.
  • Kanwal Mehreen: Lead data scientist, curating regionally relevant datasets.
  • Muhammad Arham: Research engineer, focusing on fine-tuning models for accuracy in local dialects.

Academic Affiliations:

Our team also collaborates with esteemed academics who lend their expertise to ensure our models are rigorously tested and rooted in the latest research.

  • Maaz Amjad: Professor and Linguistics researcher specializing inNLP.
  • Sabur Butt: Expert in computational linguistics and low-resource language processing.
  • Raghavan Muthuregunathan: Researcher focused on LLM architectures and optimization for underrepresented languages.

Why LARGE Matters

In a world dominated by English-language AI models, many communities have been left without access to tools that feel natural or relevant. At LARGE, we believe AI should serve everyone, no matter their language or location. By developing AI models that truly understand and respect local languages and cultures, we're making strides toward a more inclusive future.

Our models are designed not only to understand words but also the context and culture behind them. This makes them powerful tools for a wide range of applications—from education and healthcare to business and social services, enabling communities to leverage AI in ways that feel authentic and meaningful.

What’s Next?

The journey of LARGE is just beginning. We’re committed to bringing AI to the global stage, empowering non-English speaking regions with models that meet the highest performance standards while honoring the unique aspects of each language and culture. Arabic Llama is our next major step, but our ambitions go beyond that. We’re continually exploring new languages and looking to bring our technology to other communities around the world.