Carl Luis Pöhl

Clinical AI · TU Dresden

• –PhD on AI that helps doctors choose cancer treatment.

Parkinson’s progression · TU Dresden

• –Medical doctorate on predicting how Parkinson’s disease progresses.

TU Dresden

• –Final exam expected December 2026. Second state exam (M2) good (2.0), ~top 20% nationally (2025). TU Dresden since 10/2022, after pre-clinical studies at Semmelweis, Budapest. Rotations: Argentina, Namibia, Switzerland, Germany.

RWTH Aachen University

• –Grade good (2.0), part-time alongside medicine. Mathematics very good (1.3). Business Administration minor.

EPFL, Lausanne, Switzerland

• –Exchange semester funded by a SEMP scholarship.

HKUST, Hong Kong

• –GPA 3.74/4.0, Dean's List in both semesters, top 10% of cohort.

Co-founder & AI Consultant, AI Consultancy

• –Won and co-delivered AI back-office automation projects worth £15,000–£20,000 each. Manage ongoing client relationships throughout deployment.
• –Built a client’s agent infrastructure and evaluation framework for post-release quality checks.

Co-founder, Natural-Medicine AI Assistant

• –Built a live AI research assistant that checks supplements against trusted sources such as Cochrane reviews, rates how strong the evidence is and flags drug interactions and safety risks.
• –Set product priorities and coordinated the technical and marketing teams through launch.

Co-founder, AI for Cancer Treatment Decisions

• –Built multi-agent AI that helps cancer specialists identify treatment options from medical guidelines, with checks for unsupported answers and clinician oversight. Kept patient data inside hospital infrastructure.
• –Led the technical team, translating clinical advisers’ input into requirements and priorities.
• –Dresden Exists incubator · Y Combinator Spring 2026 interview

Connectome Neuroscience Society

• –Co-led neuroscience and neurosurgery seminars, workshops and conferences with 100+ attendees.
• –Managed interdisciplinary teams including physicians and pharmaceutical representatives.

Transaction, Strategy & Execution Intern

• –Co-developed a client's market-exit kick-off strategy and prepared global execution workshops.

Data Science Intern, Crop Science Division

• –Built the division's first machine-learning sales-targeting model. Backtest indicated up to ¥120M savings. Presented the findings to the head of the Asia-Pacific division.

Dept. of Urology, University Hospital Dresden

Collaborators: Jakob Kather (Kather Lab), Lars Hilgers. Supervisors: Sherif Mehralivand, Marcus Sondermann.

• –Secured four university hospitals to test the system. Wrote all ethics submissions through approval. Testing on new patients is planned from Q4 2026.
• –Benchmark. Compared 44 setups across six AI designs on 500 real patient cases kept out of development. Checked the automatic scoring against eight specialist doctors reviewing the same 50 cases.
• –Introduced a separate test of whether AI follows treatment guidelines: every system built around the guidelines did better than a model trained on past decisions, even though that model agreed with doctors in 99.2% of cases.

Dept. of Neurology, University Hospital Dresden

• –Built a model that predicts how Parkinson’s progresses from 17 routine clinical scores in 409 patients, designed to say when it is unsure instead of guessing. Checked its predictions against outcomes in a later patient group.

DKFZ and University Hospital Heidelberg

• –Developed AI that tells pancreatic cancer, inflammation and normal tissue apart in microscope images: 96.8% balanced accuracy on 475 tissue slides, always tested on patients the model had not seen.

Kather Lab (EKFZ / TU Dresden)

• –Used AI models that read DNA sequences to tell 19 cancer types apart, reaching 0.94 AUROC and 55.0% accuracy. Improved one model’s accuracy by almost 10%.