Data Science & ML Claude 技能

Arboreto GRN Inference

Infers gene regulatory networks from expression data using high-performance machine learning algorithms like GRNBoost2 and GENIE3.

LatchBio Integration

Builds and deploys serverless bioinformatics workflows using the Latch Python SDK and cloud infrastructure.

Metabolomics Workbench Database

Accesses the NIH Metabolomics Workbench to query metabolite data, study metadata, and standardized nomenclature for biomarker discovery.

Scikit-learn Machine Learning Toolkit

Implements comprehensive machine learning workflows including classification, regression, and data preprocessing using the industry-standard Scikit-learn library.

PyTorch Model CLI Implementation

Implements standalone command-line inference tools in C, C++, and Rust by extracting weights and logic from PyTorch models without Python dependencies.

PubChem Database Explorer

Accesses the world's largest chemical database to retrieve compound properties, structures, and bioactivity data for cheminformatics workflows.

pyOpenMS Proteomics Toolkit

Processes mass spectrometry data for proteomics and metabolomics analysis using the pyOpenMS library.

ChEMBL Database for Drug Discovery

Queries ChEMBL's vast database of bioactive molecules and drug discovery data for medicinal chemistry research.

DeepChem Cheminformatics

Applies machine learning to chemistry, biology, and materials science to predict molecular properties and design new compounds.

Zarr Python

Manages large-scale N-dimensional arrays with chunking and compression for high-performance scientific computing and cloud storage.

Astropy Scientific Computing

Performs advanced astronomical data analysis, coordinate transformations, and cosmological calculations using the industry-standard Astropy library.

Portfolio Optimization with C Extensions

Accelerates Python numerical computations by implementing performance-critical mathematical algorithms as high-speed C extensions.

Train FastText

Optimizes FastText text classification models by balancing hyperparameter tuning with strict file size and accuracy constraints.

Small Matrix Eigenvalue Computation

Optimizes dominant eigenvalue calculations for small dense matrices by reducing Python wrapper overhead through direct LAPACK integration.

Caffe CIFAR-10 Build and Training

Guides the compilation of the Caffe deep learning framework from source and the execution of CIFAR-10 image classification training workflows.

Caffe Framework & CIFAR-10 Training

Guides the compilation of the legacy Caffe deep learning framework and the training of convolutional neural networks on the CIFAR-10 dataset.

Count Dataset Tokens

Calculates token counts in large-scale datasets using specific tokenizers and precise filtering criteria.

Bayesian Network Fitting and Modification

Streamlines Bayesian Network workflows by guiding structure learning, parameter estimation, causal interventions, and network sampling using industry-standard libraries.

Adaptive Rejection Sampler

Implements efficient adaptive rejection sampling algorithms for generating random samples from log-concave probability distributions.

Video Processing

Guides frame-level analysis and event detection in videos using OpenCV to ensure accurate motion tracking and algorithm validation.

LLM Inference Batching Scheduler

Optimizes LLM inference workloads on compilation-based accelerators by balancing request batching, shape selection, and padding overhead to minimize costs while meeting latency requirements.

ARC-AGI Merge & Pattern Recognition

Manages the merging of conflicting git branches and the development of robust, generalized pattern recognition algorithms for ARC-AGI grid transformation tasks.

SAM Cell Segmentation

Implements advanced image segmentation pipelines using SAM and MobileSAM to extract high-precision cell boundaries and polygon coordinates from images and structured data.

Build Cython Extensions

Provides systematic guidance for building and installing Cython extension packages while resolving compatibility issues with modern Python and NumPy versions.

ReLU Network Model Extraction

Extracts hidden layer weight matrices from black-box ReLU neural networks using input-output query patterns and geometric analysis.

Torch Tensor Parallelism

Implements distributed tensor-parallel linear layers in PyTorch to enable training of models that exceed single-device memory limits.

PyHealth

Develops, tests, and deploys healthcare-specific machine learning models using electronic health records, clinical prediction tasks, and medical coding systems.

GWAS Catalog Explorer

Queries the NHGRI-EBI GWAS Catalog for genetic variants, SNP-trait associations, and summary statistics to support genetic epidemiology research.

Torch Pipeline Parallelism

Implements distributed model training by partitioning PyTorch layers across multiple GPUs using pipeline parallelism patterns like AFAB and 1F1B.

MuJoCo MJCF Tuning

Optimizes MuJoCo MJCF simulation files to improve computational performance while maintaining high physics accuracy and stability.