llm-systems-lab denis-samatov · PARTIAL
DDP/FSDP2 training and an LLM serving engine built from scratch — paged KV cache, continuous batching, prefix caching, speculative decoding, int8 — tested for equivalence and benchmarked
github.com/denis-samatov/llm-systems-lab · ★ 0 · Forks 0 · Size 614 KB
SUMMARY
Technologies 11
Scored 9
Observed 9
Practices 6
Evidence 12
Skips 2
COVERAGE
Analyzed 81 files · 4 commits · 0 API calls
TECHNOLOGIES & DEPTH
Shell LANGUAGE Depth 70
1 files · PRODUCTION
JSON LANGUAGE Depth 70
12 files · PRODUCTION
Markdown LANGUAGE Depth 70
20 files · PRODUCTION
TOML LANGUAGE Depth 70
1 files · PRODUCTION
YAML LANGUAGE Depth 70
6 files · PRODUCTION
Python LANGUAGE Depth 70
32 files · PRODUCTION
Poetry BUILD_TOOL Depth 80
1 files · CONFIGURATION
NumPy LIBRARY Depth —
0 files · config only
Pydantic LIBRARY Depth 58
1 files · PRODUCTION
Ruff QUALITY Depth —
0 files · config only
pytest TESTING Depth 49
4 files · TEST
PRACTICES
documentation · observedautomated_tests · observedcontinuous_integration · observedcontainerization · absentlinting · observedformatting · absent
ACTIVITY & OWNERSHIP
First commit 2026-10-22
Last commit 2026-10-22
Active months 1
Commits 4