skills.rest
FMurray
FMurrayCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

model-serving

Deploy and query MLflow models and GenAI agents on Databricks Model Serving endpoints.

Updated Mar 5, 2026
Search Tags:#deployment#langgraph#endpoint#mlflow#databricks#model serving#unity catalog
One-click install
npx skills add https://github.com/FMurray/mlfts --skill model-serving-fmurray
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: model-serving
Source: https://github.com/FMurray/mlfts/tree/main/.agents/skills/model-serving
Command: npx skills add https://github.com/FMurray/mlfts --skill model-serving-fmurray

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires mlflow, databricks-langchain, langgraph, databricks-agents, pydantic, databricks-sdk.

What problem does it solve? Deploying ML models and LLM agents to production endpoints involves many error-prone steps: logging models with correct signatures, registering to Unity Catalog, declaring resources for authentication, handling long async deployments, and querying endpoints in the right format. This Skill provides end-to-end guidance for the entire Databricks Model Serving lifecycle. ## Core Features & Use Cases - Classical ML Deployment: Deploy sklearn, xgboost, LightGBM, and PyTorch models using MLflow autolog with automatic Unity Catalog registration. - GenAI Agent Deployment: Build and deploy ResponsesAgent and LangGraph agents with tool integration via Unity Catalog Functions and Vector Search retrievers. - Async Job-Based Deployment: Avoid MCP timeouts on ~15-minute agent deployments by running deployment as a Databricks job. - Endpoint Querying: Query endpoints via MCP tools, Python SDK, REST API, or OpenAI-compatible clients. - Use Case: You built a LangGraph agent with UC Function tools and need to log it with proper resources, register it to Unity Catalog, deploy it to a serving endpoint, and test it with chat-style queries. ## Quick Start Ask the assistant to deploy your MLflow model or ResponsesAgent to a Databricks serving endpoint and then query it with a test message.

Frequently Asked Questions about model-serving

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I deploy an MLflow model to a Databricks serving endpoint?▼

Enable MLflow autolog with registered_model_name pointing to a Unity Catalog path, train your model, then create a serving endpoint via the Databricks UI, MLflow Deployments SDK, or Databricks SDK. Enable scale-to-zero to reduce idle costs.

How do I deploy a LangGraph agent on Databricks Model Serving?▼

Implement the agent as an MLflow 3 ResponsesAgent, log it with mlflow.pyfunc.log_model including resources and pip_requirements, register to Unity Catalog, then deploy with databricks.agents.deploy(). Deployment takes about 15 minutes, so use a job to avoid timeouts.

Why does my ResponsesAgent return an invalid output format error?▼

ResponsesAgent requires helper methods instead of raw dicts for output items. Use self.create_text_output_item(text, id) for text, create_function_call_item for tool calls, and create_function_call_output_item for tool results.

What Databricks Runtime version do I need for agent development?▼

DBR 16.1 or later is recommended because it pre-installs most GenAI packages including mlflow 3.x, langchain, and pydantic. You still need to pip install databricks-langchain, langgraph==0.3.4, and databricks-agents.

Why does my deployed agent get authentication errors when calling tools?▼

The model was logged without declaring its resources. Add DatabricksServingEndpoint, DatabricksFunction, and DatabricksVectorSearchIndex entries to the resources parameter of mlflow.pyfunc.log_model so Databricks provisions credentials automatically.

How do I query a Databricks serving endpoint from Python?▼

Use WorkspaceClient from the Databricks SDK and call w.serving_endpoints.query with the endpoint name plus messages for chat agents or dataframe_records for ML models. Endpoints are also OpenAI-compatible for use with the openai client library.

Related Skills

Explore More Software Engineering Skills→
datasciencemonkeydatasciencemonkey
27

databricks-model-serving

Deploy and query Databricks Model Serving endpoints for ML models and AI agents.

Community
Advanced
syeda-hoorain-alisyeda-hoorain-ali

llm-provider-configuration

Configure the OpenAI Agents SDK with a custom base URL and API key for alternative LLM providers.

Community
Intermediate
raphaelmansuyraphaelmansuy
2

agent-evaluation-mlflow

Implement agent evaluation and safety gates using MLflow 3.x scoring and tracing workflows.

Community
Advanced
databrickslabsdatabrickslabs
38

databricks-agent-bricks

Develop conversational AI applications on Databricks with pre-built Agent Bricks.

Official
Intermediate