Generative Programming with Mellea: from Agentic
summary
Mellea is an open‑source Python library that replaces giant prompt essays with small imperative calls, explicit instruct–validate–repair steps, and pluggable sampling strategies so nondeterminism is circumscribed and debuggable. It provides a CLI to produce adapter checkpoints, a session abstraction for backends and contexts, and component decorators that turn LLM calls into typed functions. Case studies show portability gains (one workload falls from failing on a small model to matching prior results after a rewrite, a 78% benchmark lift), and activated adapters cut re‑prefill costs.
AI-written summary · about
chapters (9)
- 0:00Generative programming intro
- 27:49Hello world with Malaya
- 32:52Instruct-validate-repair pattern
- 42:10Generative functions and checks
- 52:28Rich document processing
- 56:22Model inference design
- 1:09:37Activated LoRA checks
- 1:11:23Case study: GPT researcher prompt
- 1:23:12Summary & Q&A
related talks
15mFrom agent soup to proper software design: Mellea puts developers back in control
32mProgramming language evolution in the AI era
49mPrompt Darwinism: Evolutionary Engineering in the Age of LLMs
01h 39mMastering LLMs: From Prompt Engineering to Agentic AI
01h 59mEmerging Patterns for Coding with Generative AI
24mNatural Language Coding Revolutionising Programming with GenAI
50mFrom Prompt Hacking to Architectural Determinism: Engineering Reliable GenAI Systems
12mNavigating the World of AI Development Without Going Insane