Thanh Tung (Todd) Cao Email & Phone Number

A deep learning and computer vision practitioner and researcher with broad knowledge on image (object) search, classification and object detection. Particularly interested in reading and implementing cutting edge research papers to solve real-world problems. Experienced with leading a smaller R&D team to develop an end-to-end training pipeline with the emphasis on usability, expandability and reproducibility. My expertise also includes algorithms and computational geometry. During my PhD, I created the fastest GPU Euclidean distance transform algorithm that promises exact result. In my PhD dissertation, I developed the first series of GPU algorithms to construct several fundamental geometric structures such as convex hull and Delaunay triangulation (with collaborations from a few others). I have also implemented these algorithms with numerical stability and degeneracy tolerance, while still achieving orders of magnitude speedup over traditional sequential implementations. As a practical engineer, I always pursue novel idea that just work and can solve a wide range of daily business problems. At the same time, being a research, I love to dive down into all the experiment details to fully understand and explain the implemented solution. I also consider myself a performance engineer who loves to optimize my code, redesigning my data-structures and algorithms to get the best out of the hardware at hand. Nonetheless, sometimes I do prefer readability and bug-resistant implementation over minuscule performance gain. I am always seeking new problems to solve, be it a totally new way of doing things, or just to make existing software a lot faster.

Listed skills include Cuda, C++, Parallel Algorithms, Computational Geometry, and 7 others.