Hi, I’m Adithya. I run an open source AI research lab called Cognitivelab (cognitivelab.in), an open source first research lab focused on building inclusive, high impact AI for India and beyond. We recently received the LLaMA Impact Grant from Meta, becoming the only team from India to do so. :contentReference[oaicite:0]{index=0}

My open source work spans language models, document AI, and agentic systems. I created Omniparse (GitHub), a framework for converting unstructured data into structured formats, now with over 6,000 GitHub stars. I also led the development of Ambari (Hugging Face Collection), India’s first bilingual Kannada English LLM.

One of my proudest contributions is the Nayana Initiative, a long term effort to build a complete ecosystem of models, datasets, benchmarks, and tools for multilingual, multimodal, multitask AI across 22 Indic languages. This includes:

• NayanaOCR, a lightweight OCR model accepted at NAACL 2025 Workshops
• Nayana 1M, the largest multilingual document understanding dataset

We recently launched NetraEmbed (Hugging Face Model), a state of the art multilingual multimodal retrieval model that generates unified representations for both document images and text queries across 22 languages, enabling efficient and accurate semantic document retrieval. :contentReference[oaicite:1]{index=1}

also run the AI Engineering.Academy, where I teach applied AI concepts and share hands on tutorials for developers and students.

If you are passionate about open and accessible AI, especially for underrepresented languages, your sponsorship helps me continue building these systems, supporting research, and mentoring the next generation of AI builders. Thank you for supporting my work.