Summary
Prakamya Mishra is a research scientist specializing in large-scale LLM and generative model training, currently advancing pre-training, post-training, and evaluation workflows on AMD GenAI clusters. With nine years of experience spanning industry and academia, she has driven novel RLHF and few-shot sampling methods that improved factual consistency in clinical summarization and cross-domain keyphrase extraction, leading to publications at NeurIPS workshops and EACL. Her background includes applied research roles at Amazon and IBM, independent climate-KG and spoken-word representation projects that earned spotlight and conference acceptances, and a MS in Computer Science from UMass Amherst. Notably, she combines systems-level expertise in cluster-scale training with practical model innovations—such as multi-encoder cross-attention and tokenization strategies that yielded multi-fold gains—making her adept at turning research ideas into scalable training solutions. Based in Seattle, she maintains an active research portfolio and public presence at prakamya-mishra.github.io.
9 years of coding experience
5 years of employment as a software developer
Master of Science - MS Computer Science, Master of Science - MS Computer Science at University of Massachusetts Amherst
Bachelor of Technology Computer Science, Bachelor of Technology Computer Science at Shiv Nadar University
English, Hindi, Gujarati