Tech • IA • Crypto
Builders Unscripted: Ep. 5 - Derya Unutmaz
7/10TL;DR
A biomedical researcher is using advanced AI models to build custom tools, simulate immune systems, and push toward personalized “digital twin” medicine.
KEY POINTS
Early recognition of biology’s complexity
After completing medical training, Derya Unutmaz turned to biomedical research and quickly concluded that traditional methods were insufficient to handle biology’s scale. With trillions of components and billions of interactions, biological systems appeared too complex for conventional analysis. This realization in the early 1990s led to an early interest in artificial intelligence as a necessary tool for future breakthroughs.
AI’s turning point with reasoning models
The arrival of advanced reasoning systems, particularly in 2024, marked a decisive shift. When tested on complex immunology problems, newer models produced deeply structured, cross-domain insights that earlier systems could not. This capability signaled that AI could move beyond summarization into hypothesis generation and experimental reasoning, making it directly applicable to scientific discovery.
Daily research transformed by AI coding tools
AI-assisted coding tools such as Codex have enabled rapid development of specialized scientific software. Tasks that once required weeks or months can now be completed in hours or days. These tools are used to build applications for simulations, data analysis, and visualization, fundamentally changing the pace and accessibility of computational biology work.
Custom-built flow cytometry analysis software
One major application involves flow cytometry, a technique that analyzes thousands to hundreds of thousands of individual cells. Using AI, a fully functional analysis platform was built to process cell data, identify immune cell types such as CD4 and CD8 T cells, and generate statistical outputs. The system handles complex datasets in real time, rivaling long-established commercial tools.
Simulation of immune system behavior
AI is also being used to simulate cellular decision-making processes. A prototype simulator models T cell receptor signaling, allowing researchers to test how variables like ligand strength or inhibitory signals affect immune responses. These simulations can predict downstream molecular activity, offering a new way to explore outcomes before running physical experiments.
From static images to interactive scientific models
Image generation tools have been combined with coding systems to create interactive biological visualizations. A single AI-generated image of an immune atlas was transformed into a dynamic web-based simulator where users can explore cell interactions, manipulate variables, and observe outcomes such as the effects of PD-1 checkpoint inhibition in cancer therapy.
AI-designed CRISPR gene editing tools
AI-driven applications are also being used to design CRISPR gene-editing targets. By inputting a gene such as CD4, the system retrieves DNA sequences, identifies optimal editing sites, ranks them, and enables rapid design of gene-editing libraries. This accelerates experimental workflows and introduces customization beyond existing tools.
Toward digital twins in medicine
A central goal is the creation of digital twins—comprehensive simulations of an individual’s biology, including genetics, immune systems, and metabolism. Such models could predict disease risk, simulate treatments, and enable fully personalized therapies. This approach could dramatically reduce clinical trial timelines from years to days by testing interventions virtually.
Implications for cancer and personalized treatment
Oncology is already moving toward personalization, with treatments tailored to specific mutations. AI could extend this further by integrating immune response variability and other biological factors. Experimental approaches, including custom RNA vaccines, suggest a future where therapies are designed for individual patients rather than broad populations.
Automation and the future of scientific research
The scientific process itself is expected to change. AI systems may generate hypotheses, simulate experiments, and analyze results in continuous loops, while automated labs execute physical tests. Researchers would shift toward guiding objectives rather than manually conducting each step, accelerating discovery across biology, chemistry, and physics.
Cultural resistance and rapid progress
Despite skepticism in parts of the scientific community, rapid improvements between model generations have demonstrated exponential progress. Advanced systems are now capable of producing insights comparable to decades of human experience, suggesting a narrowing gap between human intuition and machine reasoning.
CONCLUSION
AI is rapidly transforming biomedical research from slow, manual experimentation into a high-speed, simulation-driven discipline, with digital twins and fully personalized medicine emerging as the next frontier.