Shailza Jolly

Hi! My name is Shailza Jolly. I am an ML Scientist / AI Engineer with a Ph.D. in Computer Science and 6+ years spanning academic research and industry. My work centres on making LLM systems accurate and auditable enough to trust: retrieval and grounding, agentic systems, and training-data quality at scale.

I completed my Ph.D. at TU Kaiserslautern, where I worked with my advisor Prof. Andreas Dengel, focusing on developing machine learning methods for building natural language generation and understanding systems in data-constrained settings. My research interests include vision and language systems, interpretability, and conversational AI.

I completed my Master’s in Computer Science from TU Kaiserslautern and spent a semester abroad at Kyushu University in Japan, where my work How do Convolutional Neural Networks Learn Design? won the Best Student Paper Award at ICPR 2018. I was also named AI Newcomer of the Year 2021 by the German Informatics Society and BMBF. I completed my Bachelor’s from GNDEC, India.

During my studies and career, I have worked at Amazon AI, NVIDIA Research, Amazon Alexa, SAP AI Research, and the German Research Center for Artificial Intelligence (DFKI).

I have been a visiting researcher in the labs of Prof. Seiichi Uchida, Prof. Lili Mou, and Prof. Isabelle Augenstein.

Over the past year I’ve been building independently, going deep on the parts of the modern LLM stack that matter most for reliability: retrieval and grounding, agentic system design, and inference efficiency. Sift is the clearest example, an open-source hybrid retrieval system whose agent answers only from retrieved passages and cites the source page for every claim. I also built an agentic tutoring system with persistent memory and voice that I use daily, and wrote about attention, data quality for LLM training, and the shift from generative to agentic AI.

I’m now looking for my next full-time role as an ML Scientist / AI Engineer, in Germany or remote within the EU. My research background lets me get up to speed on hard problems quickly, my engineering work means I write the code and ship it, and I’m at my best translating between technical and business goals, mapping product KPIs to the metrics that move them.