Guides UMAP workflows for 2D/3D embeddings, clustering preprocessing, supervised UMAP, and Parametric/Aligned UMAP.
A skill that guides Hugging Face Transformers work: Hub model loading, pipeline inference, text generation, and Trainer fine-tuning.
Turn a gene symbol, genomic region, or FASTA into promoter, splice, enhancer, chromatin, expression, and gene-annotation predictions via hosted DNA language models.
Guides Claude in scaling pandas/NumPy workloads beyond RAM using Dask DataFrames, Arrays, Bags, Futures, and schedulers.
Gives Claude expert use of datamol, the Pythonic RDKit wrapper, for SMILES parsing, standardization, fingerprints, clustering and 3D conformers.
Restructures PyTorch code into LightningModules and configures multi-GPU training, logging, and checkpointing.
Safely detects the CPU, memory, disk, scheduler and accelerator limits that actually apply to the current process, producing a redacted JSON snapshot plus conservative workload plans.
Discovers and wires up curated Hugging Face datasets, models, and Spaces across 17 scientific domains, from protein design to climate modeling.
Teaches Claude Code to build, deploy and scale Python and AI/ML workloads on Modal's serverless GPU cloud.
Builds auditable market research reports where every claim maps to a source and TAM/SAM/SOM plus forecasts are presented as scenarios.
Teaches Claude idiomatic Polars — expressions, lazy queries, and pandas-to-Polars migration.
Lets Claude drive the pyzotero client to search, create, update and export your Zotero reference library.
Describe a scientific diagram in plain language and get a publication-oriented PNG that an AI reviewer scores against your document type's quality bar.
Gives Claude fine-grained RDKit guidance for SMILES/SDF parsing, descriptors, fingerprints, substructure search, reactions and 2D/3D generation.
Writes competitive research proposals tailored to NSF, NIH, DOE, DARPA, and Taiwan NSTC review criteria, formats, and budget rules.
Zero-shot time series forecasting with Google's TimesFM foundation model — no training required.
Systematically evaluates research claims for design flaws, bias, statistical errors and evidence quality using GRADE and Cochrane Risk of Bias.
A reference skill that makes Claude write modern, correct seaborn 0.13 code for distributions, relationships and categorical comparisons from pandas DataFrames.
Guides Claude through rigorous statsmodels work — OLS, GLM, discrete choice, and time series — with full diagnostics and inference.
Query U.S. national debt, federal spending, Treasury auctions and exchange rates through the free Fiscal Data API — no key required.