AI Code Optimization Assistant
Publicada el 2026-07-16
Descripción de la oferta
I’m building an in-house AI agent whose single focus is to review my Python and JavaScript projects and return advanced optimisation advice—think full-scale refactors, algorithmic efficiency gains, and performance-focused design changes, not just cosmetic tweaks. Here is what I need from you: • A self-contained agent (script, service, or lightweight web UI—whichever best suits rapid iteration) that accepts a codebase or select files, analyses them, and returns structured recommendations covering complexity reduction, execution speed, memory usage, and modularity. • Explanations must be actionable: highlight current bottlenecks, show improved snippets, and justify each change so I can learn from the process. • The model or pipeline should handle real-world projects of moderate size without timing out or flooding me with irrelevant lint-style warnings. • Solutions leveraging open-source LLMs (e.g., GPT-J, LLaMA 2, Code Llama) or a clever orchestration of existing APIs are welcome, as long as latency stays practical and privacy is respected; any fine-tuning steps should be reproducible. • Include a straightforward way to update the knowledge base so the agent improves as our code standards evolve. Acceptance criteria: the agent runs locally on macOS/Linux, delivers optimisation reports for both languages in under three minutes on a 5 k-line project, and its recommendations measurably reduce runtime or complexity in at least one provided sample. I’m ready to supply test repositories and benchmarks as soon as we start. Let me know your preferred tech stack and any clarifications you need to hit these goals.
Skills
Fuente original: freelancer