Omprakash JenaEnterprise AI Architect

Global aerospace engine manufacturer

AI project charter generation

Manufacturing · Project management

Production

Rule-driven complexity scoring plus LLM generation. 30–40 hours to under 2 minutes; 300+ users.

My role
AI Solution Architect
Delivery
Team of 7
Status
In production

<2 min

to generate a charter, down from 30–40 effort hours

Problem

Writing a project charter took a project manager 30–40 hours: gathering stakeholder input, interpreting governance requirements and compiling the documentation. Charters often arrived incomplete or inaccurate, which led to governance-review delays, rework and deferred approvals.

Approach

Project managers answer a set of questions about the project and upload the governance documents that apply. The system scores the project's complexity, then generates a governance-compliant charter grounded in those documents and in historical charters. It also recommends the skill sets the project needs from a skill-mapping repository, and lists the governance requirements that apply at that level of complexity. The project manager reviews and refines the draft before it goes for approval.

Architecture

Deployment view. Project managers on the corporate network sign in with Microsoft Entra ID and reach Azure Front Door, which serves a React single-page app from a Storage static website. The app calls API Management, which routes to a LangChain charter service on AKS. The service reads historical charters from Azure Database for PostgreSQL, caches access tokens in Azure Cache for Redis, fetches tokens on refresh from an access management service, writes logs to Azure Log Analytics and stores charters in a Storage account. In a Databricks AI platform workspace it generates the charter through Model Serving, which calls Azure OpenAI LLM and embedding models, searches skill mappings in Vector Search and sends traces to MLflow. Uploaded governance documents from the service and historical charters from PostgreSQL flow into a bronze, silver and gold medallion in a separate Databricks data workspace; gold data is embedded and indexed for Vector Search and feeds Power BI dashboards, which project managers view and the app embeds.

Outcome

A complete first draft now takes under 2 minutes instead of 30–40 hours of manual effort, for 300+ users, with higher charter quality, better governance compliance and faster project starts.

Technology stack

  • Azure Front Door
  • Azure Storage
  • API Management
  • AKS
  • Azure Database for PostgreSQL
  • Azure Cache for Redis
  • Azure Log Analytics
  • Azure Databricks
  • Databricks Model Serving
  • Databricks Vector Search
  • MLflow
  • Azure OpenAI
  • Power BI
  • Microsoft Entra ID
  • LangChain
  • React

Client and project names are withheld. Architecture and outcomes are described in general terms to respect client confidentiality.