Data & Analytics

Convert per-instance RLE or polygon annotations to COCO JSON format for seamless use with Detectron2 and MMDetection

Extract per-frame CNN features then classify the temporal sequence with stacked GRU layers and a boolean mask for variable-length video inputs

Uses a binary classifier's probability to gate object detector outputs in three tiers: keep detections, append a no-finding box, or replace all detections.

Memory-efficient Keras generator that streams sharded CSV files in chunks, renders strokes to images on-the-fly, and yields batches for training on datasets too large for memory

Interpolate missing bounding boxes across video frames using bidirectional pandas interpolation to maintain smooth tracking through occlusions

Wrap albumentations bbox transforms in a 10-attempt retry loop that re-rolls the augmentation when all boxes get cropped out, preventing empty-target samples from poisoning the detection loss with NaN gradients

Applies albumentations augmentations to object detection data while preserving bbox-label correspondence via BboxParams.

Applies Mixup augmentation to 3D volumetric images and their segmentation masks, interpolating both inputs and loss targets.

Specialist skill for Python data engineering — pandas, polars, DuckDB, numpy, ETL pipelines, tabular data ingestion, and notebook-to-module extraction. Use when working with dataframes, data validation at ingress boundaries, merge/join operations, typed column contracts, or choosing between pandas vs polars vs DuckDB for a data task.

Statistics, probability, linear algebra, and mathematical foundations for data science

Supervised & unsupervised learning, scikit-learn, XGBoost, model evaluation, feature engineering for production ML

Analyze Ethereum transactions including details, status, gas usage, and internal transfers

Office document processing - PDF, Excel, CSV, Word, YAML, JSON, INI, TOML, XML

Parse and analyze common log formats (nginx, syslog, application)

Analyze image technical details: dimensions, format, EXIF metadata, file size

Read Grafana dashboards, panels, and alerts

具有智能自动路由的统一搜索技能。使用多信号分析自动在 Serper (Google)、Tavily (研究)、Exa (神经)、You.com (RAG/实时) 和 SearXNG (隐私/自托管) 之间进行选择，并进行置信度评分。

数据可视化专家，涵盖图表选择、色彩理论和标注最佳实践。支持图表类型（柱状图、折线图、散点图、热力图）、坐标轴规则和数据叙事。用于：图表、图形、仪表板、报告、演示文稿、信息图、数据故事。触发词：data visualization, chart, graph, data chart, bar chart, line chart, scatter plot, data viz, visualization, dashboard chart, infographic data, data presentation, chart design, plot, heatmap, pie chart alternative

Structures data quality assessment with completeness, accuracy, and consistency validation. Use when auditing clinical data, assessing data quality, or validating data integrity.

Selects appropriate imaging protocols based on clinical indication and patient factors. Use when choosing imaging protocols, selecting contrast parameters, or determining appropriate study type.

Maps between clinical terminologies (ICD-10, SNOMED CT, LOINC, RxNorm) with semantic equivalence validation. Use when mapping medical codes, converting between terminologies, or validating code mappings.

Structures population health analysis with risk stratification and care gap identification. Use when analyzing population health, stratifying patient risk, or identifying care gaps.

Structures health data governance programs with stewardship roles, policies, and data quality standards. Use when establishing data governance, defining data stewardship, or managing data policies.

Applies GCP/ICH principles to clinical research operations with compliance monitoring. Use when ensuring GCP compliance, training on ICH guidelines, or auditing research practices.