gavinr
16 years of experience developing quantitative systems in cross-platform modern C++. The bulk of my experience has been on teams of computer vision scientists where I've converted research prototypes into production software that can run in real-time. I utilize the latest C++ standards and enjoy adapting legacy code to use modern best practices. I use Python for prototyping, algorithm exploration, utility development, and as a scripting language on top of C++.
I also have 6 years of experience as a tech lead, where I was responsible for the technical direction and architecture of a large computer vision system on a team of between 3 and 15 developers and scientists.
The most important metrics I strive for when writing code are: correctness, performance, and maintainability. I favor simple solutions and avoid complexity unless it provides value. I can help with development, performance optimization, legacy code maintenance and modernization, build system integration, and code reviews.
Algorithms include: Structure from Motion, image/video processing, 2D and 3D facial recognition, image stitching, object detection and tracking, Monte Carlo simulation, nonlinear optimization, integer programming.
C++ stack: C++23 and below, CMake, vcpkg, clang, gcc, Visual Studio, Jenkins, CPack, Linux, Windows.
Libraries: OpenCV, Ceres Solver, Eigen, OR-tools, ffmpeg, boost, STL
UI: Qt, Dear ImGui
https://www.linkedin.com/in/gavinrosenbush/
grosenbush at gmail dot com
They're suggesting that no overtaking prevents the bottleneck in the first place, not that it gets you out of an existing one.
Seems like this would prevent drivers behind overtakers from having to slow down which propagates back causing a bottleneck.
How about multi-lane highways with tolls for each lane priced dynamically based on traffic conditions.
Yes! Sounds like you've got the right mindset...good luck!