Hugging Face plugins

Design and deploy LLM applications with LangGraph agents, RAG systems, vector search, and advanced prompt engineering. Covers embedding model selection, hybrid retrieval, evaluation, and production scaling for chatbots and AI assistants.

Build and automate end-to-end ML pipelines including data ingestion, feature engineering, model training, deployment, and monitoring, with multi-agent orchestration across cloud platforms and production ML infrastructure.

Instantly access 149 ready-to-use agent skills for scientific research, data analysis, machine learning, bioinformatics, and technical writing across domains. Execute experiments, analyze sequences, run simulations, generate reports, and automate lab workflows without writing boilerplate.

End-to-end AI and data engineering suite covering data pipelines, ML model development, LLM architecture, MLOps, database optimization, and prompt engineering for production systems.

Run structured AI product evaluation and operations workflows including A/B testing, LLM agent benchmarking, product analytics, and observability.

Manage Hugging Face Hub resources—models, datasets, Spaces, jobs—and deploy, fine-tune, and evaluate ML models on SageMaker or cloud GPUs, all from Claude Code.

Manage the complete LLM fine-tuning lifecycle with eval-gated checkpoints: from data preparation and method selection (SFT, DPO, GRPO) through training with LoRA/QLoRA, vision SFT, and quantized export, all gated by capability drift and arena evaluations.

De-identify PHI, extract clinical named entities, map to FHIR R4 and OMOP CDM, run compliance audits, and evaluate model fairness—all on-device.

End-to-end SageMaker AI model lifecycle management: validate datasets, select and fine-tune models (SFT, DPO, RLVR, RLAIF), deploy to SageMaker endpoints or Bedrock, diagnose HyperPod cluster and NCCL failures, and evaluate model quality — all through natural-language instructions.

Handles visual AI tasks including image processing, OCR, barcode detection, and document analysis using latest vision models (GPT-4V, Claude Vision) with cost optimization

Implements AI/ML features including language model integration, recommendation systems, computer vision, and intelligent automation with prompt engineering, ML pipelines, and model deployment

Implement AI/ML features including LLM integration, recommendation systems, computer vision, and intelligent automation with practical prompt engineering, ML pipelines, and model deployment.

Orchestrate medical research workflows: intake projects, discover meta-analysis topics, audit study designs, generate reproducible analyses, check manuscript compliance with reporting guidelines, and prepare submission-ready documents.

Implements AI/ML features in applications: integrates LLMs, builds recommendation engines, adds computer vision search, and automates intelligent workflows with prompt engineering, ML pipelines, and model deployment.

Build and optimize RAG pipelines with document chunking, embedding generation, and vector retrieval, while writing, debugging, and optimizing LLM prompts using advanced techniques like chain-of-thought and few-shot learning.

Process images, perform OCR, detect barcodes, and analyze documents using cutting-edge vision models (GPT-4V, Claude Vision) with automatic cost optimization.

Generate deployable, GPU-accelerated Vision AI pipelines using YAML configs and NVIDIA backends (DeepStream, Triton, vLLM, TensorRT-LLM, PyTorch) for high-performance video and streaming inference microservices.

Scans repositories to inventory ML models, training pipelines, data sources, feature pipelines, and monitoring, answering 'what ML do we have' queries for model inventory and ML assessment.

End-to-end ML/AI engineering — builds ML pipelines from feature engineering to baseline training, evaluates model performance with drift detection, designs LLM integrations (RAG, agents, fine-tuning), and creates production prompt packages with evals.

Fine-tune LLMs with PEFT/LoRA, RLHF, and instruction tuning, plus optimize prompts using few-shot and chain-of-thought techniques, with built-in auditing for quality gaps

Search code and files with ripgrep and fd, transform JSON/YAML with jq/yq, generate diagrams from Mermaid/D2 text, process images with ImageMagick, and automate shell workflows — all from Claude Code.

Train, evaluate, export, and deploy NVIDIA TAO computer vision models — covering classification, detection, segmentation, pose estimation, depth, video understanding, and 3D perception — with integrated AutoML, DEFT iterative improvement loops, synthetic data generation, and multi-platform GPU job submission (Docker, Kubernetes, SLURM, Brev, DGX Cloud).

Debug ML training failures (OOM, NaN, divergence), validate hyperparameters and configs against official docs before runs, and generate grounded implementation plans for fine-tuning, RAG, and inference pipelines using a knowledge base of 27k+ pages from Hugging Face, PyTorch, DeepSpeed, and vLLM.

Designs and evaluates retrieval reranking pipelines with relevance scoring, learning-to-rank models, and NDCG/MRR metrics for search and RAG systems.

Run a virtual AI university in Claude Code that teaches AI/ML, robotics, medicine, and ethics through structured lectures, practical tasks, and progress tracking across departments

Deploy vLLM inference servers on Docker, Kubernetes, or bare metal and benchmark their performance (throughput, latency, TTFT, TPOT) using synthetic or real datasets, including prefix caching tests.

Search, compare, run, and prompt AI models hosted on Replicate, including deploying custom models via Cog and managing predictions with webhooks and streaming.

Install, manage, and use the Kokoro TTS engine on Apple Silicon macOS. Provides an HTTP server with an OpenAI-compatible endpoint for text-to-speech synthesis, health checks, diagnostics, and real-time audio playback.

Design a complete fine-tuning pipeline for LLMs, including PEFT configuration, dataset formatting, training loop setup, and evaluation criteria to adapt models for specific tasks.

Generate, edit, and animate images and videos using ComfyUI workflows directly from Claude Code — build text-to-image/video pipelines, run parameter sweeps, diagnose failures, install custom nodes, and visualize workflow graphs.

Run local GGUF models via Mozilla Llamafile with an OpenAI-compatible API, handling installation, server startup, GPU/CPU configuration, SDK integration, and connection troubleshooting for offline or air-gapped AI workflows.

Manage the full computer vision dataset lifecycle in FiftyOne: import, curate, deduplicate, visualize, run model inference, evaluate predictions, and export to standard formats. Also build custom plugins and UIs for the FiftyOne platform.

Design ranking pipelines with reranker selection, score fusion, and cross-encoder patterns, optimizing search or recommendation ranking while managing latency trade-offs

Automatically detect a running FreeRide gateway and route any OpenAI-compatible client through it to access free-tier AI inference from providers like OpenRouter, Groq, NVIDIA NIM, Cloudflare Workers AI, and HuggingFace.

Enforce a falsification-first research pipeline for empirical ML claims: bootstrap hypotheses, preregister experiments, reproduce baselines, run adversarial falsification, and gate decisions (kill-or-ship) with statistical rigor and evidence verification.

Select and optimize vision models (GPT-4V, Claude Vision, Mistral-OCR) for OCR, barcode detection, document processing, and multi-modal AI workflows, balancing speed, accuracy, and cost.

Orchestrate an extreme programming agent framework with multi-agent loops, Rust/Python skill datums, MCP servers, and governance gates for autonomous task decomposition, code review, and continuous improvement across AWS, GitHub, and AI model providers.

Index PDFs, markdown, and source code into Qdrant for semantic (vector) and MeiliSearch for full-text (keyword) search, then query both with the arc CLI. Supports AST-aware chunking, frontmatter, and git metadata.

Generate agent teams with authentic Korean-language personas from the Nemotron-Personas-Korea dataset, including workplace honorifics, speech levels, and generational/regional diversity for any domain.

Reproduce Long Video Sparse Attention (LVSA) paper headline results including SotA grid comparisons and latency scaling using bundled benchmarks, VQeval, and VBench-Long scoring, with figure regeneration.

Improve LLM apps and agents from real production traces: capture traces, build evals, run local optimization (GEPA), compare models/providers, and route through the Understudy gateway to reduce cost/latency and raise quality/reliability.

Coordinates a fleet of 144+ specialized AI agents across memory, governance, research, content, energy, marine, music, and cosmos domains with persistent vaults, cross-repo orchestration, and sovereign policy enforcement.

Develop dApps on Ritual Chain with skills for smart contracts, async precompiles, React/Next.js frontends, on-chain AI/ML, and automated debugging agents.

Generate and run synthetic Korean survey respondents from the Nemotron-Personas-Korea dataset — load personas by UUID or demographics, dispatch batched surveys with parallel fan-out, and validate synthetic distributions against real human benchmarks using Cohen's w and chi-square metrics.

Search, browse, and transcribe Swedish National Archives (Riksarkivet) records using HTR transcription, archival search, document viewers, and Label Studio annotation workflows.

Frames data science problems, preprocesses and validates data, performs EDA, designs and runs ML experiments with reproducibility checks, evaluates models, assesses deployment readiness, and compounds learnings into structured knowledge base.

Design, build, and debug dynamic neural networks that grow, prune, or adapt topology during training. Includes lifecycle state machines for neural modules, gradient isolation, PEFT/LoRA adapters, and modular composition. Expert advisor diagnoses growth and pruning issues via code search and logs.

Deploy and manage AI/ML workloads on RunPod cloud GPUs, including serverless inference, training, pod provisioning, and infrastructure selection across providers like AWS and Hugging Face.

Design, train, fine-tune, and deploy production-grade LLMs and NLP systems with expert guidance on model architecture, prompt engineering, RAG, inference optimization, and MLOps infrastructure.

Discover, evaluate, and acquire datasets from multiple sources (Kaggle, HuggingFace, IPFS, arXiv, DBLP) for AI model training and fine-tuning. Assess quality, licensing, and provenance, then download free or paid data directly.

Provides 139 scientific skills for computational biology, drug discovery, quantum computing, and research workflows. Includes tools for protein engineering, single-cell genomics, molecular ML, bioinformatics database queries, literature review, manuscript writing, and lab automation.

Implement AI/ML features using an agent specialized in LLM integration, recommendation systems, computer vision, and intelligent automation. Handles prompt engineering, ML pipelines, and model deployment.

Expert in vision models, OCR, barcode detection, and document analysis using GPT-4V, Claude Vision, and latest visual AI models with cost optimization.

Accelerate scientific research with 148 skills spanning bioinformatics, cheminformatics, quantum computing, ML, data analysis, lab automation, literature review, and grant writing

Implement AI/ML features: integrate LLMs, build recommendation systems, computer vision, and intelligent automation with prompt engineering and model deployment

Handles image processing, OCR, barcode detection, and document analysis using vision models like GPT-4V and Claude Vision with cost optimization

Implement AI/ML features, integrate language models, build recommendation systems, and add intelligent automation to applications.

Automate visual AI tasks—image processing, OCR, barcode detection, and document analysis—using cutting-edge vision models (GPT-4V, Claude Vision) with cost optimization.

Orchestrate end-to-end ML pipelines with automated data preparation, model training, hyperparameter tuning, deployment, and monitoring. Use autonomous agents for advanced analytics, production ML systems, and infrastructure management across cloud platforms.

Build production-ready LLM applications with LangGraph agents, RAG systems, vector search, and prompt engineering. Covers embedding model selection, hybrid search fusion, evaluation metrics, and vector index tuning for scalable AI systems.

Build and manage dora-rs robotic dataflow applications: configure YAML dataflows, develop Rust/Python nodes, integrate ML/vision/audio pipelines, control robots, record datasets for imitation learning, and debug with log analysis and Mermaid graph generation.