Research tools & benchmarks
Python research workflows and C++ performance work built around reproducible experiments and comparable results.
- Status
- Recorded work
- Type
- Independent software projects
- Period
- Project archive
- Location
- Independent research setup
- My role
- Software development & analysis
The work
I built Python tools for data acquisition, caching, structured logging, scenario comparisons, tests and report generation. AlphaFoundry separates research, portfolio construction, validation and execution simulation into modules.
A measured example
In a separate C++ sparse matrix-vector multiplication benchmark, I improved recorded throughput from approximately 0.28 to 1.3 GFLOP/s, about 4.6 times the baseline. The work used OpenMP and memory-access optimization, including cache-aware reasoning and prefetching.
Why the structure matters
The inputs, baseline, change and test conditions need to remain visible. Reproducible workflows make it possible to revisit an experiment and compare the result with the next version.
Independent research code and benchmark records. The GitHub link leads to my public repository collection.
The benchmark is specific to its test conditions. Financial-data experiments are research, not a record of investment returns.
Research and software on GitHub