I'm an AI/ML researcher at MILab, North South University's privately funded AI lab, working on alternative architectures, long context, and reasoning. I came to research from venture, after two years in early-stage investing at Bangladesh Angels. I also work on AI at Zelf and run the NSU AI Community .
Mustavi Khan
AI researcher in Dhaka. Architectures, long context, reasoning.
Recent posts
See all posts →Reading highlights
See all →- Mamba: Linear-Time Sequence Modeling with Selective State Spaces Gu & Dao · 2023
The selective state space model that made an attention-free architecture competitive with Transformers.
- Lost in the Middle: How Language Models Use Long Contexts Liu et al. (TACL 2024) · 2023
Models use information best at the start and end of context, and worst in the middle.
- Training LLMs to Reason in a Continuous Latent Space (Coconut) Hao et al. (FAIR / UCSD) · 2024
Reasons in latent space by feeding the last hidden state back as the next input embedding.
Let's connect
Reach out about research, building, or community — or just to say hi.