hi, i’m niels
n (.) muendler (at) posteo (.) de
I research Code LLMs
Recent LLMs are remarkably strong at generating functionally correct code. Yet, they still generate unsafe code and lack reliability. My research focuses on measuring and improving LLM code generation in these direction. Check out my papers to learn more!
I started in 2017 at TUM, spent the second half of 2019 (“fall”) at NUS. I graduated as M.Sc. CS at ETH in 2024 and am now a PhD Student under Martin Vechev.
I like working with Isabelle/HOL, LLMs and Python (usually these fields do not overlap). If you are interested in my CS related stuff, head over to my github page. If you are interested in my research, check out my research page. You can also find me on Twitter
Other topics
BüKE, EJM and Haydee! e.V are all great projects/institutions I recommend checking out.