„NextSilicon” produces intelligent accelerators for parallel operations. A regular program can be compiled with „NextSilicon”’s proprietary compiler, and the binary will use the accelerator hardware, implemented as an ordinary PCIe card, in order to compute whatever can be calculated in parallel. This gives up great increase in the speed of calculations, which is useful for physics applications and for molecular biology applications, — for example, calculating the tertiary and quaternary structures of proteins.
Actually, one can say that „NextSilicon” produces a very specialized and sophisticated mathematical coprocessor.
Work Experience
- Started on January 27, 2025, and currently continue working there.
- Position: Senior Embedded Software Engineer.
My Achievements
- Created a decision tree-based system that helps analyzing unexpected reboots of the supercomputer servers.
- Set up CI/CD pipeline for regression and iterative testing in Jenkins.
- Took part in developing the test suites and utilities.
- As a part of System Validation team, I’ve developed code that prevents hangs or deadlocks in the accelerator’s management program.
- I found and eliminated several thread-unsafe places where critical data could be corrupted, resulting in unexpected behavior.
- Developed code for the Zephyr RTOS, updated the device tree and managed hardware resources.