Hi, this is Nahid. I am an independent researcher with Cohere Labs community, working on Multimodal Learning, Computer Vision and Embodied AI.

I recently created Maya – a multilingual multimodal LLM. I work at the intersection of multimodal learning, computer vision and embodied ai, developing models that perceive, reason, and act in the physical world.
My current interests include:

• Spatial understanding in VLMs
• Physics-aware world models
• Multimodal Learning
• Causal Learning

Publications

• The Spatial Blindspot of Vision-Language Models. Nahid Alam et al. pre-print arXiv

• Behind Maya: Building a Multilingual Vision-Language Model.
  Nahid Alam et al. CVPR 2025 Workshop (VLMs4All).
  arXiv · Google Scholar

• Understanding and Mitigating Toxicity in Image-Text Pretraining Datasets: A Case Study on LLaVA.
  Karthik Reddy Kanjula, Surya Guthikonda, Nahid Alam, Shayekh Islam.
  CVPR 2025 Workshop (ReGenAI), Oral.
  arXiv · Google Scholar

• Embedding Geometries of Contrastive Language-Image Pre-Training.
  Jason Chuan-Chih Chou, Nahid Alam. ECCV 2024 Workshop (Beyond Euclidean).
  arXiv · Google Scholar

More at Google Scholar

Recent Projects

• Maya: Multilingual multimodal foundation model (2 CVPR workshops)
• Gemma3n-VLA: Vision-Language-Action model built with Hugging Face LeRobot
• GR00T-N1 Hackathon: Bimanual robot manipulation with multimodal control

🌐 Connect

• LinkedIn: nahidalam
• Twitter / X: @nahidalam