Virtual AI Project Coach System
In-progressPythonTypeScriptNext.js+5 more

Virtual AI Project Coach System

AI-powered project coaching platform featuring end-to-end system architecture, requirement analysis, and full-stack implementation.

Timeline

2024 – Present

Role

Full-Stack Developer & System Architect

Team

Agile Engineering Team

Status
In-progress

Technology Stack

Python
TypeScript
Next.js
React
PostgreSQL
Tailwind CSS
Prompt Engineering
Agile Methodology

Key Challenges

  • System Architecture Planning
  • Multi-tier Context Preservation
  • AI Hallucination Mitigation
  • Dynamic Requirement Elicitation
  • Agile Team Coordination

Key Learnings

  • End-to-End System Design
  • Scalable Microservices Architecture
  • Prompt Engineering & Guardrails
  • Relational Schema Optimization
  • Academic Evaluation & Defense

Virtual AI Project Coach System

Overview

The Virtual AI Project Coach System is an advanced AI-guided platform developed as our university graduation capstone project at OSTİM Technical University in Ankara, Turkey. Designed to act as an automated, intelligent technical advisor for computer engineering students and software teams, the system bridges the gap between academic project ideas and structured, production-ready deliverables.

As the Full-Stack Developer & System Architect, I led the foundational architectural planning, requirement analysis, database schema modeling, and full-stack web implementation.


The Problem Statement

During university capstone and software engineering projects, student teams consistently encounter common pitfalls:

  • Abstract Requirement Scoping: Difficulty translating vague initial project ideas into concrete, testable functional specifications.
  • Delayed Academic Feedback: Supervisors often have dozens of teams to advise, creating bottlenecks when students need urgent architectural guidance.
  • Unstructured Milestones: Teams jumping straight to coding without proper system design or relational modeling, resulting in costly rewrites.

The Virtual AI Project Coach solves this by providing automated, domain-aware mentorship at each stage of the engineering lifecycle.


System Architecture

We designed a decoupled, multi-tier system architecture engineered for reliability, modularity, and rapid iteration:

+-----------------------------------------------------------+
|              Presentation Layer (Next.js 14)              |
|   - Dynamic Kanban & Milestone Boards                     |
|   - Interactive Architecture Diagrams                     |
|   - AI Chat & Prompt Interface                            |
+-----------------------------------------------------------+
                             |
                             v  (RESTful JSON / SSE)
+-----------------------------------------------------------+
|               API & Business Logic (Node.js)              |
|   - User Authentication & Team Workspaces                 |
|   - Milestone Validation & State Machine                  |
|   - Audit Logging & Versioning                            |
+-----------------------------------------------------------+
                             |
                             v
+-----------------------------------------------------------+
|          AI & Prompt Engineering Engine (Python)          |
|   - Context Memory & Requirement Extraction               |
|   - Domain-Specific Prompt Templates                      |
|   - Verification & Anti-Hallucination Guardrails          |
+-----------------------------------------------------------+
                             |
                             v
+-----------------------------------------------------------+
|                Database Layer (PostgreSQL)                |
|   - Normalized Relational Schemas                         |
|   - Indexed Project Trees & Versioned Artifacts           |
+-----------------------------------------------------------+

Key Features & Capabilities

  • Automated Requirement Elicitation: Guides students step-by-step through user stories, functional requirements, and non-functional constraints.
  • Architectural Trade-Off Analysis: Evaluates proposed tech stacks (e.g., SQL vs. NoSQL, microservices vs. monolith) based on project scale and team skills.
  • Milestone & Sprint Management: Integrates sprint planning tools with AI milestone verification, highlighting blockers before deadlines.
  • Automated Technical Documentation: Generates initial software requirement specifications (SRS) and architecture overviews in markdown and PDF formats.

Full-Stack Implementation Details

Frontend Architecture

  • Built with Next.js 14, React, and TypeScript, leveraging modern component patterns and responsive Tailwind CSS layouts.
  • Real-time updates for team collaborations and streaming AI responses using server-sent events (SSE).

Backend & AI Pipelines

  • Python-powered AI Orchestrator: Manages model prompts, enforcing strict system prompts and structured JSON outputs so responses can be programmatically rendered.
  • PostgreSQL Database: Configured with strict foreign key constraints, connection pooling, and optimized indexing for rapid milestone retrieval.

Agile Methodology & Academic Defense

  • Two-Week Sprints: Conducted daily standups, backlog grooming, and sprint review demos.
  • Rigorous Git Workflows: Managed code branching, issue tracking, and peer-reviewed pull requests.
  • University Evaluation Presentation: Delivered formal presentations demonstrating live AI coaching sessions and received commendations from the faculty evaluation panel.

Future Roadmap

  • GitHub Repository Webhook Integration: Automatically analyzing pull requests against defined milestone requirements.
  • Multi-Agent Simulation: Implementing specialized AI agents (e.g., Security Advisor, QA Architect, Database Optimizer) collaborating in real time.
  • Broader Academic Deployment: Expanding the platform across multiple engineering departments at OSTİM Technical University.