First, clarify the filter's intended purpose and constraints (e.g., real-time, embedded, safety-critical). Then systematically review the code for correctness, efficiency, and robustness issues, explaining the impact of each deficiency and suggesting improvements.
Pro tip: Prioritize deficiencies by severity and relate them to Zipline's domain (e.g., real-time drone delivery), showing you understand business context. Mention that even minor bugs can have safety implications.
Ask about the filter's expected behavior, input/output, performance constraints, and deployment environment. This ensures your analysis is relevant and targeted.
Check for algorithmic errors: incorrect filter coefficients, boundary handling, initialization, and numerical stability. Verify against expected output for test cases.
Analyze time and space complexity, memory access patterns, and potential optimizations (e.g., loop unrolling, SIMD, circular buffers). Consider real-time constraints.
Look for missing error handling, input validation, overflow/underflow risks, and thread-safety issues. Consider edge cases like empty input or extreme values.
List deficiencies by severity, explain their impact, and propose fixes. Highlight any trade-offs between fixes.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.