| # Marco Lopez — ML Engineer | |
| NLP and applied machine learning. Current work: controllable summarization (salient-information extraction + LLM steering), efficient local inference on Apple Silicon (MLX), gesture recognition from skeletal time series. | |
| - GitHub: [LookUpMark](https://github.com/LookUpMark) | |
| - Site: [lookupmark.github.io](https://lookupmark.github.io) | |
| Featured work: | |
| - [SM-SIP](https://huggingface.co/collections/LookUpMark/sm-sip-salient-information-extraction-and-prompting-6a9b041bde695a232959affb) — salient-information extractors for LLM-steered summarization (English + Italian, Longformer-based) | |
| - [Ornith-1.5 · MLX Quantizations](https://huggingface.co/collections/LookUpMark/ornith-15-mlx-quantizations-6a9b0c25fac182a7a2d8a847) — 4/6/8-bit MTP quants for Apple Silicon (oMLX) | |
| - [DYLEM-GRID](https://huggingface.co/collections/LookUpMark/dylem-grid-dynamic-gesture-recognition-6a9b042a7689303ec8ed8157) — dynamic hand-gesture dataset (Leap Motion, 400 recordings) | |