How krishnatiwari005/datapilot-ai works
DataPilot AI is an Agno-based, multi-agent assistant for data-science and machine-learning work. A user request enters through the AgentOS application layer and is coordinated by the Data Science Team, which delegates to specialist agents for CSV loading, dataframe exploration, visualization, coding, file management, and execution. The documented workflow extends through cleaning, feature engineering, model training, and evaluation, with user review before generated code is saved and run. Agent definitions and wiring are in app.py; unsampled implementation details are represented only where README or sampled instructions support them.
Main entry points: app.py Data Science Team · car_details.csv Sample CSV
Components 19
Application interface
- AgentOS / API
app.py
Agent coordination
- Data Science Team
app.py - Data Loader Agent
app.py - File Manager Agent
app.py - Data Understanding Agent
app.py - Visualization Agent
app.py - Coding Agent
app.py - Shell Agent
app.py
Data science workflow
- CSV tools
app.py - Dataframe operations
app.py - Visualization tools
app.py - ML code workflow
app.py
Project runtime
- Project files
- Sample CSV
data/car_details.csv - Python execution
app.py - Technical web search
app.py - SQLite session memory
app.py
Other components
- User
- Groq model service
Connections 35
- User → AgentOS / API: sends request
- AgentOS / API → Data Science Team: serves team
- Data Science Team → Data Loader Agent: delegates loading
- Data Science Team → File Manager Agent: delegates files
- Data Science Team → Data Understanding Agent: delegates analysis
- Data Science Team → Visualization Agent: delegates plots
- Data Science Team → Coding Agent: delegates coding
- Data Science Team → Shell Agent: delegates execution
- Data Loader Agent → CSV tools: uses tools
- CSV tools → Sample CSV: reads CSV
- Data Loader Agent → Project files: searches files
- File Manager Agent → Project files: manages files
- Data Understanding Agent → Dataframe operations: calls operations
- Dataframe operations → Sample CSV: loads CSV
- Visualization Agent → Dataframe operations: inspects data
- Visualization Agent → Visualization tools: creates plots
- Visualization Agent → Project files: uses files
- Coding Agent → ML code workflow: develops pipeline
- Coding Agent → Technical web search: researches
- Coding Agent → Python execution: uses Python tools
- Coding Agent → User: requests code review
- Shell Agent → Python execution: runs project code
- Shell Agent → Project files: inspects files
- Data Science Team → SQLite session memory: stores session context
- Data Loader Agent → SQLite session memory: uses memory
- Data Understanding Agent → SQLite session memory: uses memory
- Visualization Agent → SQLite session memory: uses memory
- Coding Agent → SQLite session memory: uses memory
- Shell Agent → SQLite session memory: uses memory
- Data Science Team → Groq model service: uses model
- Data Loader Agent → Groq model service: uses model
- Data Understanding Agent → Groq model service: uses model
- Visualization Agent → Groq model service: uses model
- Coding Agent → Groq model service: uses model
- Shell Agent → Groq model service: uses model