Ankit Bhagatwala's Email Addresses & Phone Numbers

💼 Past Experience

Work experience at TeslaAnkit Bhagatwala is currently a Staff Machine Learning Engineer, Autopilot at Tesla in Palo Alto, CA.Played key role in migrating the vision inference stack from pipeline based to a graph based scheduler. This has allowed multi-camera and other specialized neural networks to be run on Tesla's AI chip without sacrificing performance. This was achieved while constantly shipping customer releases on multiple vehicle platforms Code refactors to improve robustness, platform independence and error handlingIntroduced tools for better monitoring of distributed, multi-task neural network training frameworkPerformance optimization of distributed training platform and framework while maintaining training accuracy (collaborative effort with Nvidia and PyTorch devs).Senior Research Scientist (Machine Learning) at Pilot AI from August 2016 to August 2018 (2 years 1 month) in Palo Alto, CA.As team lead for performance optimization, developed high-performance deep learning algorithms that beat current state-of-the-art detection systems on heavily compute-constrained ARM devices in speed while maintaining accuracy. Implemented fast convolution and other math kernels that achieved 40% speedup over leading open source libraries for embedded platformsHelped build a fully homegrown deep learning training and inference framework (similar to PyTorch/TensorFlow) for embedded vision. Primary contribution was in the automatic differentiation (autodiff/autograd) engine and adding distributed data parallel capability using MPI on the training cluster. As one of the early employees, built automated systems for testing, modeling and deployment from the ground up.Computer Systems Engineer/HPC Consultant at Berkeley Lab from September 2015 to May 2016 (9 months) in Berkeley, CA. Implemented capabilities for editing code, compiling and submitting batch jobs to compute platform through web browser for the NERSC user portalMyNERSCTeam of three postdocs in achieving a 2x speed up of core kernels of the WARP particle accelerator code through increased memory locality and vectorization.(Among 15 students from North America chosen by NSF to attend an intensive two-week GPU programming workshop in Valparaiso, Chile.)

🎓 Education

Ankit Bhagatwala's education didn't just begin with a high school diploma or a degree from a school in India. It started with plenty of hard work and dedication lasting all the way up until his Noblership, in 2011.Ankit Bhagatwala's education started in 2000 with a B.Tech. at the Indian Institute of Technology, Madras. From there he moved on to M.Tech. at the same institute in 2005. After completing his M.Tech. in 2005, he decided to move back to India to do his Ph.D. in Aerospace Engineering at Stanford University.After completing his Ph.D. in Aerospace Engineering at Stanford University, in 2011, Ankitbhagatwala decided to move to the US to take up a Fellowship, Data Science at The Data Incubator. It was at The Data Incubator where he started to learn all about the data analytics process and how to apply it in his day-to-day work.Now, having completed over 20 years of education and work experience, Ankitbhagatwala is a senior fellow at The Data Incubator, and aiologist and cardiologist, respectively. He is also the author of two books, "Data Analytics: A Byte-By-Byte Guide" and "Into the Data Age: how machines are transforming our lives and businesses."

💡 Technical & Interpersonal Skills

Ankit Bhagatwala has extensively played a key role in the development of severalaska machine learning and Neural networks related algorithms, concepts and tools, as well as ensuring their robustness and performance. He has also been involved in the development of blockchain-based solutions and has written multiple papers on the topic. Additionally, he has also developed various tools for optimization, including a widely used widely used deep learning machine learning library, as well as a performance analysis tool for large subspace neural networks.