Skill UI

This skill guides the setup and execution of GPU-intensive Machine Learning workloads on Modal, a serverless cloud platform. It details a zero-config workflow for training, fine-tuning, and inference, eliminating the need for manual SSH setup or complex Docker orchestration. Key steps include detailed cost estimation, selecting the optimal GPU based on VRAM requirements, and generating reproducible Python launcher scripts for reliable ML experimentation.

PathML is a comprehensive Python framework designed for computational pathology workflows. It supports end-to-end analysis of Whole Slide Images (WSI), including loading 160+ proprietary formats, performing advanced stain normalization and tissue segmentation, constructing spatial cell graphs, and training deep learning models (e.g., HoVer-Net) for multiplex immunofluorescence and spatial proteomics. Ideal for large-scale biomedical image analysis.

A comprehensive tool for creating high-quality training datasets for custom object detection and segmentation models. It supports multiple advanced annotation methods including BBox drawing, SAM2 pixel-perfect segmentation, and DINOv3 visual grounding. Features include AI auto-detection, object tracking, and standardized COCO format export, streamlining the entire data pipeline from capture to training.

A powerful command-line interface (CLI) tool, modeled after kubectl/docker, designed for comprehensive management of GPU compute jobs on the Qizhi platform. It supports job submission (interactive/non-interactive), resource discovery, node availability checking, and batch processing for distributed training and HPC workloads. Ideal for machine learning engineers and data scientists running large-scale scientific computations.

This skill simulates a Chief Data Officer's deep interrogation, pressure-testing any plan touching core data strategy. Use it before approving ML model training runs, selecting data infrastructure (warehouse vs. mesh), productizing data assets, or making major team hires. It ensures that all data decisions are grounded in clear business decisions, compliance, and structural necessity.