Performance optimization with GitHub Copilot

Software Development with GitHub Copilot

Thalia Barrera

AI Engineering Curriculum Manager, DataCamp

The silent slowdown

nanobanana: half: A split illustration contrasting a fast responsive application shown as a smooth rocket launching on one side and a slow spinning loading indicator on the other, flat vector style

Performance issues:

Don't show up in unit tests
Don't cause immediate crashes

📈 Code looks correct, but doesn't scale

A bottleneck in the wild

def find_common(list1, list2):
    common = []
    for item in list1:
        for item2 in list2:
            if item == item2:
                common.append(item)
    return common

Code works fine for small inputs
1 million items → 1 trillion comparisons
$O(n^2)$ algorithm!

❌ Brings down production on a busy day

[screencast]

Tell Copilot what to optimize for

nanobanana: half: A performance dashboard with speedometer dials and metric gauges showing different optimization indicators, flat vector illustration on a clean background

Speed → faster algorithms, caching, batching
Memory → resource cleanup, streaming, generators
Both → full performance review

Performance prompts that work

Targeted optimization

Optimize #selection for performance and reduce time complexity

Review #selection for memory leaks

Codebase-wide scan

@workspace find functions with nested loops over large collections

[screencast]

Bake performance into every suggestion

.github/copilot-instructions.md

## Performance standards
- Prefer set or dict lookups over nested loops
- Use context managers for files, connections, and sockets
- Avoid loading entire datasets into memory, use generators or streaming

✅ Prevents performance bottlenecks from being written in the code

Let's practice!

Software Development with GitHub Copilot