Nadja Flechner

I'm a researcher with a background in AI. My goal is to use my skills to benefit the public and help non-profit organizations. My current focus is on forecasting methodologies at the Forecasting Research Institute (FRI), where I'm exploring how forecasting can enhance decision-making processes. I'm also diving deeper into the capabilities and implications of Large Language Models (LLMs). I will attend ARENA to further develop my technical skills LLM research.

I've also worked on the intersection of AI and remote sensing — developing weakly supervised learning models that extract meaningful insights from satellite imagery. I'm excited about the potential of remote sensing to deliver positive societal impact.

Research Interests

These are some topics and questions I'm interested in, and hoping to research more:

• Where can AI create meaningful efficiency gains in research?
• How can we ensure AI developments proceed safely and responsibly?
• How can forecasting techniques be leveraged for better institutional decision making?
• What impacts will increased AI adoption have on societal discourse, democratic processes, and public opinion formation?
• Which applications of remote sensing can deliver positive societal impact?

Research Services

I hope to be of use for the non-profit sector, and for organizations that are looking to leverage AI and research to achieve their goals. Based on my background, I can help with the following services:

• Data analysis and visualization
• Quantitative research
• Automating processes with AI
• Finding ways to make your work more efficient
• Literature reviews

Work Experience

At the Forecasting Research Institute, I contribute to research projects, and have implemented LLM-based solutions to increase the efficiency of their research.

Continuing work developed for my MSc thesis, where I developed a weakly supervised model to automatically segment permafrost land structures from aerial images. This work enables efficient tracking of changes in permafrost areas in northern Scandinavia, a process previously done mostly manually.

Researched the applicability of generative adversarial networks (GANs) for cloud cover prediction, specifically nowcasting. Conducted this work as a 6-month independent research project, implementing solutions using PyTorch and cloud computing technologies.

Worked at an impact venture capital firm investing in sustainable start-ups and scale-ups. As part of the 'Green Industries' team, I handled incoming dealflow, conducted technology deep-dives, and analyzed technological, financial, and market risks/opportunities for various startups.

As a TA for Artificial Intelligence and Data Science courses, I wrote lesson plans, created course materials, developed the course syllabus, and taught classes.

Assisted PhD students with their research during data collection and interview phases (transcribing), as well as document formatting and proofreading.

Education

Grade: Cum Laude
Thesis: Developed a weakly supervised segmentation model for the Ecology and Environmental Science (EMG) department at Umeå University. Grade: 9/10.
Internship: Conducted a 6-month deep learning research project in meteorology at Infoplaza, developing a cloud nowcasting model. Grade: 9/10.
Skills: Deep Learning, Machine Learning

Grade: Summa Cum Laude
Pursued an interdisciplinary education with focus on Artificial Intelligence. Combined courses in programming, mathematics, psychology, and philosophy to build a strong foundation for advanced studies in AI.

Outside of Work

When not doing research, I enjoy exploring new places through travel and spending time outdoors - often in the mountains doing alpine sports (climbing, hiking, splitboarding). I also enjoy reading, and surroudning myself with people that have interesting and diverse opinions.

Contact

Email: {my first name}flechner99@gmail.com