Merge pull request #12 from yusrmuttaqien/dev/dataset-generator-almer

refactor: simplify CV parser to send raw files directly to LLM vision endpoint

feat(llm-client): optimize LLM calls with batching and combined methods

feat: add LLM availability status to health check endpoint

feat: menambahkan generator dataset versi bersih (scripts/dataset_generator.py dan file pendukungnya src/generator/)

feat: add /parse-cv endpoint for extracting student data from resumes

feat: add /recommend-single endpoint for single scholarship matching

docs: clarify LLM service requires manual enable in API docs

feat: add LLM-powered recommendation text generation

feat: add scholarship metadata export and fit scores to API

Update landing page subtitle to include synthetic data generation

feat: add API base URL display with copy button

feat: gracefully handle missing HF token and add API landing page

feat: add Docker support for HuggingFace Spaces deployment

feat: add HuggingFace artifact sync for data and model repos

refactor(config): consolidate serving config into default.yaml and add retraining support

Merge pull request #6 from yusrmuttaqien/dev/fusion-almer

feat(serving): add cached embedding loading and restructure export

fix: membenarkan masalah seed operasi-operasi kecil

fix(api): improve response models and add field descriptions

refactor: replace chronological train/val split with optional holdout fraction

feat(serving): add configurable retraining data splits

chore(serving): destructure serving layer and related shared codes

refactor: extract training logic into dedicated training loop module

feat(training): add model retraining with new CSV data support

feat: add tensorboard monitoring and improve precompute staleness detection

refactor: simplify serving config and consolidate configuration handling

docs: add FastAPI serving and update project documentation

add project files