Updated Coding Experts – Software Engineering
Q1 (Programming Best Practices)
Ranking: C > B > A.
Model A contains an indentation bug that can prevent correct execution and also uses integer
division that truncates even-length medians. Model B is correct and readable but more verbose
than necessary. Model C is the strongest because it correctly handles empty lists, avoids mutating
the input, and cleanly computes the median for both odd and even lengths.
ES
Q2 (Describe for Reconstruction)
Decode strings encoded as k[encoded_string], where k is a positive integer that may contain
multiple digits. Nested encodings are allowed, and each decoded substring is repeated exactly k
C
times. Characters outside brackets are copied directly to the output. An empty input returns an
VI
empty string.
R
Q3 (Algorithmic Expansion)
SE
The gas station circuit problem first checks feasibility by ensuring the total gas is at least the total
cost across all stations. If the cumulative balance becomes negative while simulating from a
candidate start, none of the stations traversed so far can be valid starting points, so the next station
becomes the new candidate. A single pass is sufficient to identify the correct start. When a solution
E
exists, the valid starting index is guaranteed to be unique.
R
Q4 (Systems Design Principles)
A
Ranking: B > C > E > D > A (Reliability > Maintainability > Availability > Consistency > Scalability).
W
Reliability is the highest priority because a system that fails unpredictably cannot deliver value.
Maintainability follows because long-term evolution, bug fixing, and feature delivery depend on
FT
clean, understandable code. Availability, consistency, and scalability remain important but build
upon a reliable, maintainable foundation.
SO
Q5 (Reward Hacking)
A model could game the rubric by copying evaluation criteria or repository descriptions verbatim to
satisfy criteria 5 and 7 without demonstrating genuine reasoning. It could also inflate its response
LE
with vague, low-specificity statements to appear comprehensive and satisfy criterion 4 while
avoiding direct violations of criterion 6. A robust rubric should therefore reward evidence-based
analysis and penalize generic filler.
C
Q6 (Quality Assurance — Sudoku Repositories)
A
(a) 12-point rubric:
R
1. README clearly explains installation and usage.
O
2. Code is modular and well organized.
3. Functions/classes have meaningful names.
4. Automated tests are included.
5. Documentation explains project structure.
6. Git history shows meaningful commits. [DB]
7. Beginner-friendly setup instructions.
8. Appropriate algorithmic complexity for beginners.
9. Responsive or usable interface where applicable.
10. Repository includes an explicit license.
11. Recent maintenance or issue activity. [DB]
12. Evaluation is based on technical quality rather than popularity alone. [DB]
(b) Recommended repository: TN1ck/super-sudoku.
Reason: Modern, well-documented, beginner-friendly architecture with a strong user experience.
HoDoKu is a powerful but specialized Java solver with a steeper learning curve, while Sudoku
Studio is a research-oriented solving library that is narrower and more advanced than needed for
beginners.
ES
C
VI
R
SE
E
R
A
W
FT
SO
LE
C
A
R
O
O
R
A
C
LE
SO
FT
W
A
R
E
SE
R
VI
C
ES