Barcelona, Catalonia, Spain
I apply model-based mathematical methods to solve complex problems in various domains, such as healthcare, automotive, renewable energies, and society. With 8 years as a Research Scientist at IBM Research, a Ph.D. in Electrical and Computer Engineering from the University of California, Santa Barbara, and a Dipl.-Ing. in Engineering Cybernetics from the University of Magdeburg, I have over twelve years of research experience in both academia and industry. My core competencies include control engineering (my PhD's emphasis), physical modelling, parameter estimation, computer vision, and reinforcement learning. I have published over 20 journal and conference papers and filed more than 10 patents in these fields. My most recent achievement was leading and delivering a three-year research project that combined these skills and technologies to detect cancer using fluorescence data from endoscopic videos. This project involved collaborating with medical experts and delivering a novel and impactful solution.
Seeing how the sausage is made.
At IBM I have ▷ developed algorithms to track regions of interest on tissue throughout endoscopic videos ▷ developed parametric models and fit them to fluorescence data collected during colonoscopies ▷ developed machine learning algorithms that distinguish cancerous from benign growths based on those models ▷ led the IBM work package of the government-funded project involving the previous three bullet points, which mostly meant overseeing adequate and compliant reporting ▷ developed theory and algorithms for machine learning/ parameter estimation of decision makers in systems where the decision might be influenced by recommenders/ secondary decision makers ▷ implemented such algorithms in Matlab and Python ▷ implemented OpenCV and dlib-based image processing on a RaspberryPi for in-car applications ▷ contributed to project coordination and dissemination activities ▷ led and contributed to well over 20 publications in peer-reviewed conferences and journals ▷ submitted 20+ patent applications, 10 of which have been granted, with the remaining ones filed and pending
Research into distributed-parameter systems, maintenance of Rijke Tube experiment, general assistance
Researched control of distributed-parameter systems, using frequency domain methods. Worked on optimal periodic control using Fourier methods, convex optimization/cvx, sum-of-squares methods/ SOStools, and polynomial homotopy continuation. Derived distributed-parameter model of Rijke tube/thermoacoustic instability from first principles; built Rijke tube experiment, implemented system identification and real-time control using Matlab/Simulink Real Time Workshop and analog circuitry.
Developed algorithm identifying signaling pathways in human cells using evolutionary optimization methods, graph theoretic tools and binary logic. Integrated fast vectorized implementations and visualization using GraphViz in Matlab Toolbox CellNetOptimizer (CNO).