NU
Nikhil UdgataTechnical Project Manager
Enterprise Architecture Case StudyMulti-Role Real-Time Cockpit & AI Assistant

Vault β€” Multi-Role Project Delivery Cockpit & AI Assistant

Vault is an enterprise-grade, multi-role project management cockpit designed for modern engineering teams. It bridges the gap between client visibility, engineering execution, and program management by uniting structured intake, resource capacity planning, and bi-directional Jira Cloud syncing into a single, real-time workspace.

Capacity SafetyReal-Time Alerts>8h/day overload badges
Jira Sync Speed< 1 SecondBi-directional Webhook Sync
AI Token Efficiency-90% API FeesRedis 4-hour response TTL
State BrokerRedis SSE TunnelZero page-reload live updates

πŸ—ΊοΈ Delivery Journey & System Architecture

Vault aligns four organizational roles into a continuous delivery loop, synchronized in real-time via a Redis Pub/Sub Server-Sent Events (SSE) tunnel and an incoming Jira webhook stream:

[1. Client Intake & RICE] ──> [2. TPM Capacity Allocation] ──> [3. AE Execution & Standup] ──> [4. Admin Cockpit & AI Analytics]
         β–²                                                                                                    β”‚
         └────────────────────────── Bi-Directional Jira & Webhook Loop β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

Technical Stack Overview

  • FrontendReact SPA, Vite, Tailwind CSS V3, SVG Gantt charting engine, SSE hooks.
  • BackendNode.js, Express, Jira Cloud REST APIs, Redis Pub/Sub broker, PostgreSQL.
  • AI OrchestrationDify Docker stack (Ollama / Llama 3), lean-prompting, 4h Redis TTL cache.

πŸ–ΌοΈ User Portals & Visual Walkthrough

Role 01 β€” Client Partner

1. Client Portal β€” Prioritized Intake & Timeline

Clients gain a clear, scoped view of their active milestones and can submit delivery requests, scoring them mathematically using the RICE model (Reach, Impact, Confidence, Effort).

Client Dashboard β€” RICE Model Prioritized Intake
Role 02 β€” Technical Program Manager

2. TPM Workbench β€” Drag-and-Drop Capacity Scheduling

Technical Program Managers (TPMs) schedule sprints by dragging Jira tickets directly onto engineers' calendars. The board displays real-time overload safety badgesif an engineer's allocation exceeds 8 hours a day.

TPM Capacity Board β€” Drag-and-Drop Calendar Scheduling
Role 03 β€” Account Engineer

3. AE Workspace β€” Clean Inbox & Standup Compliance

Account Engineers (AEs) land on a focused list of today's allocations. They submit their EOD standups and write status transition comments that sync directly back to the Jira issue history.

AE Inbox β€” Task Allocations & Jira Webhook Sync
Role 04 β€” Executive Leadership

4. Admin Cockpit β€” Executive Summary & Portfolio Health

Admins and leadership view cross-organizational heatmaps, override project RAG (Red-Amber-Green) signals, and consume AI-generated executive summaries cached efficiently in Redis.

Admin Dashboard β€” Portfolio Heatmaps & AI Executive Summaries

⚑ Key Synchronization Engines

Server-Sent Events (SSE) Sync

  1. When a TPM assigns a ticket, backend writes allocation and publishes to pm:events Redis channel.
  2. The Express SSE server catches the Redis event and pushes it to connected browsers.
  3. Frontend usePmEvents() hook intercepts payload, debounces by 300ms, and refreshes component state.

Bi-Directional Jira Webhook Loop

  1. When a ticket status changes in Jira, a webhook is fired to POST /api/jira/webhook.
  2. Vault records the transition, alerts the assigned engineer's Inbox, and prompts them for audit comments.
  3. Engineer explanations are posted directly back to Jira as issue comments via jira.addComment().