#dailyreport #python #ai #assistent #aiassistent #llm
I solved task for job to review Python code that verify
AI assistent model. This AI model can process text and
return some category and action, can execute action, can
learn and predict, have three ONNX submodels for content
analyzing, performance predicting and routing
optimizing.
Validation metrics is score that calculated as average
successful inferences and average latency.
Validity of tasts is hardcoded as thresholds.
There are implemented cross-component validation and
stress testing.
Stress test uses min throughput, max latency, error rate,
cpu usage, max memory usage.
All tests accept test date.
I outlined code structure, execution chains.
I suggested ways to improve code, suggest to implement
integration with browser, calendar and email of users.
I made this in one day.
I didn't get offer. I think I am too smart and not
trustfull. No one likes smart asses.
蠡