Hi, I'm Aflah, a research software engineer at MPI-SWS. I specialize in backend engineering, NLP, and deep learning, with a keen interest in applying NLP for social good. I am also interested in high speed inference, large scale distributed training and squeezing out performance on limited hardware. Currently, my research involves evaluating latent knowledge and memorization in large language models. I have previously worked on hate speech detection, and leveraging NLP to address various societal challenges as a part of undegrad thesis and related research work.
Open to researcher/research engineer/backend engineer roles
- Paper Reading Discussion: Lookback Lens: Detecting and Mitigating Contextual Hallucinations
- Adapters | Exploring VGLM Via Parameter-Efficient Transfer Learning | PEFT Methods
- Adapters | Simple, Scalable Adaptation for Neural Machine Translation | PEFT Methods
- Adapters | Parameter-Efficient Transfer Learning for NLP | PEFT Methods
- Trying to Learn Triton | P-4 (Not a Tutorial)
