Rajasekhar Josyula
Photons in, decisions out.
00 // THE JOURNEY
For the last decade, I have been hand-crafting low-level systems, calibrating complex camera sensors at factory scale, and deploying safety-critical deep learning models. My work lives at the boundary where physical photons meet real-time robotic perception.
Currently, I am a Staff Software Engineer at Tesla AI. I lead sensor bring-up, real-time camera calibration, and low-level C++ capture pipelines. I also train and deploy the neural networks used for driver monitoring (DMS) and cabin safety—shipping features that protect millions of cars and Optimus humanoid robots daily.
01 // SYSTEM PHILOSOPHY
When designing software for edge robotics, latency is not just a feature—it is safety. I build sub-10ms camera-to-NN perception loops using deterministic C++20, SIMD vectorization, and optimized CUDA/Triton kernels.
I believe in end-to-end alignment: understanding how a Sony IMX sensor behaves under temperature changes is just as important as knowing how to prune a Transformer model or balance a multi-task loss frontier.
02 // CORE INTERESTS
Beyond production autopilot systems, I spend my time exploring:
- Spatial Computing: client-side Graph SLAM, 3D Gaussian Splatting, and state estimation pipelines.
- Systems Security: zero-knowledge cryptographic relays, secure token exchanges, and network protocols.
- Low-Level Engineering: high-frequency RTOS scheduling (QNX/Zephyr) and bare-metal embedded compiler optimization.
03 // UPLINK
I am always open to discussing spatial computing, robotics, and low-level performance optimization. Reach out through any of the channels below:
- qxlsz@duck.com
- GitHub
- github.com/qxlsz
- in/rajasekharjosyula
- X (DM)
- @qxlsz
- Signal
- qxlsz.6174
- Matrix
- @qxlsz:matrix.org