About me

I am a senior applied scientist at Amazon, Berlin. I work in three areas:

• Search agents & LLMs. I’m building search agents end to end, from search tools and harness optimization to fine-tuning LLMs with agentic RL. I also helped build the auto-optimize feature in Amazon OpenSearch Service.
• DevOps & forecasting. I worked on inferring application topology from time-series and AWS configuration data, and helped build Chronos, a foundation model for time-series forecasting.
• Hyperparameter optimization & AutoML. My coauthors and I won the Best Paper Award at the first AutoML Conference (2022). We shipped features on AWS SageMaker, including Automatic Model Tuning and AutoPilot.

Before Amazon, I earned my PhD at Aalto University under Prof. Juho Rousu. We built machine-learning methods for small-molecule identification, and our method (Team Brouard) won the CASMI challenge 2016 (Category 2). My thesis was named the best Finnish bioinformatics PhD thesis of 2016–2017.