TSFT-RAG
Task-specific full fine-tuning for evidence-grounded RAG behavior across open model families.
Read more →Local RAG
Retrieval architectures for privacy-sensitive higher education, research and organizational knowledge.
Read more →LLM Evaluation
Benchmarks for grounded answers, hard negatives, structured output, citations and robustness.
Read more →XML Transformation
Hybrid rule-based and language-model pipelines for complex structured document conversion.
Read more →