YC S24

• Researching applications of various NLP and CV techniques to drastically enhance well-being and productivity of teams.• Working at the Center for Collective Intelligence — an interdisciplinary lab at the intersection of management, psychology and AI.

• Stepped down from my active role and moved to the highest-level management organ of TUM.ai, the Board.• Helping shape the long-term strategy and continuity of TUM.ai.

TUM.ai is the leading AI student initiative in Germany with 200+ active members and six-figure annual income. During my leadership at TUM.ai, we have:• Started several projects directed at educating people in AI at all levels of skill, including a basic coding program, the AI Academy, and the new R&D Department.• Successfully launched the first ever incubator for AI startups in Munich: the AI E-Lab.• Reworked internal processes to ensure smooth collaboration between the 10 departments of TUM.ai, empowered team leads to achieve more through better training, and established effective medium-term planning.

• Led an interdisciplinary team of 18 doing consultancy-style AI Industry Projects.• Increased TUM.ai revenue by 650% and laid the foundations for expanding to €100k+ in annual income by completely revamping our approach to sales, operation, and marketing.• Using TUM.ai's marketing channels, achieved a 6x increase in number of applicants, resulting in a 9% acceptance rate and a drastic improvement in applicant quality.• Involved in consulting partners to find use cases for AI, sales communications, and recruiting teams with skill sets tailored to the project.

• Helped organize fast-paced end-to-end AI projects with industry partners.• Empowered participants to gain the practical AI skills needed to kickstart their career.

• Utilized generative AI to empower everyone at Rohde & Schwarz to get insights from thousands of documents describing complex processes and equipment.• Developed a framework for evaluating the performance of SOTA LLMs including GPT-like models, LLaMA, Falcon, SantaCoder, StarCoder on various internal tasks.• Deployed LLMs on the internal NVIDIA A100 cluster to ensure sensitive data never leaves the company.

• Built classical and time series ML models to give parents accurate insights on their baby's health.• Used babies' vital metrics (pulse, respiratory rate etc.) collected by a wearable device developed at Leevi as input to the models.• Conducted an extensive research of medical publications and interviewed multiple experts.Organized this project as a member of TUM.ai.

• Scraped data from the three most popular real estate websites in Germany.• Contributed to an AI model to accurately predict the development of housing prices based on 30+ scraped parameters, empowering Heimkapital to make accurate data-driven investment decisions.Organized this project as a member of TUM.ai.