clinical-decision-support
Prepare research, evaluation and governance documents for clinical decision support, not for patient care
53k GitHub starsupdated Oct 2, 2026
Loading…
Skill xây pipeline AI từ dữ liệu tới RAG/agent, đánh giá, triển khai, và giải pháp AI theo từng ngành.
Prepare research, evaluation and governance documents for clinical decision support, not for patient care
Build and validate clinical ML pipelines with PyHealth on EHR, signals and imaging
Run zero-shot time-series forecasting with Google TimesFM, including quantiles and evaluation
Build production Apache Airflow DAGs with operators, sensors, testing and deployment
Validate data quality with Great Expectations, dbt tests and data contracts
Prepare and validate datasets for fine-tuning language models (SFT, DPO) before a training run
Select and optimize embedding models for semantic search and RAG
Combine vector and keyword search to improve recall for RAG and search engines
Evaluate LLM app quality with automated metrics, human review and LLM-as-judge
Build MLOps pipelines from data prep and training to validation and deployment
Build RAG systems for Q&A over your own documents with vector databases and semantic search
Design recommendation and ranking pipelines with the six-stage Source-to-SideEffect pattern
Turn graded eval traces and production logs into SFT examples and DPO preference pairs
Parse an onboarding packet into structured KYC fields to feed the rules engine
Apply the KYC/AML rules grid to a parsed record, rate risk and flag what needs escalation
Design AI agent architectures or audit flaws in an existing agent's architecture
Clean credit data and screen variables before building loan risk models
Build an automated evaluation pipeline for Python LLM apps and produce an improvement plan
Build and run evaluators for AI/LLM applications using Phoenix
Build and operate Python Azure IoT Edge modules with manifests, reliability and security
Design multi-agent workflows with pattern choice, handoff contracts, failure handling and cost control
Classify risk, plan conformity assessment and track obligations under the EU AI Act
Cut LLM API costs with model routing, caching and prompt compression
Design RAG pipelines: pick chunking, embeddings, vector DB and measure retrieval quality
Build LLM apps over your own data with LlamaIndex: ingest documents, index and query
Run academic benchmarks like MMLU, GSM8K and HumanEval to compare model quality
Track ML experiments, manage a model registry and deploy models with MLflow
Add runtime safety rails to LLM apps with NVIDIA NeMo Guardrails
Process large ML datasets with Ray Data, from batch inference to distributed preprocessing
Serve LLMs at high throughput with vLLM, OpenAI-compatible endpoints and quantization
Design ML pipeline infrastructure with MLflow, Kubeflow, Airflow and Feast
Plan a SageMaker model deployment: pick the inference pathway, check quotas, then hand off to the deployment skills
Train or fine-tune models with TRL or Unsloth on Hugging Face Jobs, with GGUF export
Use subagents, todo planning and human approval in Deep Agents
Build agent evals with Harbor: task specs, environments, verifiers and World Knowledge Skills
Build RAG systems with LangChain: loaders, text splitting, embeddings and vector stores
Pause LangGraph for human approval with interrupt and Command, plus 4-tier error handling
Analyze Amazon reviews to find complaints, unmet needs and competitor weaknesses
Evaluate and improve an AI agent's quality with MLflow datasets, scorers and tracing
Add MLflow Tracing to Python or TypeScript code to observe agents and LLM apps
Build demand forecasting models and improve sales forecast accuracy
Optimize grid operations: power flow, unit commitment and supply-demand balance
Answer specific questions about a loaded contract with verbatim clause citations
Apply the NIST AI RMF to advise on, plan governance for or assess an AI system