Health Data Copilot

AI Data Automation•Product Consultant•2026

An automated clinical insight generation system powered by an Orchestrator-led agentic workflow (5 specialized agents), delivering Bangla responses from complex health datasets with end-to-end tracing.

Situation

Two-fold challenge: (1) Ground-level health workers lacked consumable clinical insights from 300,000+ records for immediate action, and (2) Policy stakeholders lacked ground-truth simulations for evidence-based recommendations.

Task

Develop an automated Orchestrator Workflow (Orchestrator, Intent Specialist, SQL Specialist, Data Interpreter, and Policy Analyst) providing rapid Bangla summarization for field workers while enabling complex policy simulation for stakeholders.

Result

Successfully deployed a multi-agent system providing sub-second mobile summaries and data-backed policy suggestions with full cost-per-insight transparency.

My Contribution

•Designed 5-agent Orchestrator Workflow
•Implemented SQL Schema-on-demand mapping
•Integrated Opik OTLP cost tracing

Key Results

2 Weeks

Pilot Development Time

Developed the pilot with 5 agents-

1. Orchestrator

Multi-step Execution & Handoffs

Manages the multi-step execution logic and deterministic agent handoffs.

2. Intent Specialist

Bangla & Dialect Parsing

Processes broken Bangla and rural dialects into structured JSON intents.

3. SQL Specialist

Schema-on-demand Query Generation

Generates and executes safe, read-only PostgreSQL queries using a Schema-on-demand approach.

4. Data Interpreter

Clinical Risk & Trend Analysis

Analyzes clinical values to detect risks (e.g., pre-eclampsia) and identifies health trends for field workers.

5. Policy Analyst

WHO/SDG Benchmark Simulation

Integrates ground-level data with WHO/SDG international benchmarks to simulate outcomes for policy stakeholders.

Skills & Technologies

mastraagentic-workflowobservabilityopikbangla-summarizationhealth-datasql-automation