WorkExperio

AI‑Driven Project Management & Team Formation

Team: ARPITA V · CHANDANA B · DERIN GIRISH D G · KAMBHAM SAI SATHVIKA

FastAPI React PostgreSQL MongoDB NLP WebSocket LLM JWT Docker

📋 Abstract

The increasing need for industry‑ready graduates highlights the importance of practical, structured project‑based learning, yet traditional academic environments often lack role clarity, balanced team formation, and guided task workflows. This project aims to address these gaps by developing WorkExperio, an AI‑driven collaborative platform that automates team formation, generates customized project ideas, assigns role‑specific tasks, and provides real‑time interaction through chat and an intelligent assistant.

The system integrates NLP‑based resume parsing, clustering‑based skill matching, and LLM‑powered project/task generation, supported by a FastAPI backend and WebSocket‑enabled communication. Experimental evaluation demonstrates high accuracy in skill extraction and role prediction.

🎯 Problem Statement

Students struggle to gain real‑world project experience due to a lack of structured teamwork, role clarity, and practical collaboration, leaving them unprepared for industry roles. This project proposes an AI‑powered simulation platform that assigns students to teams, provides role‑specific tasks, and enables collaborative project execution.

❌ Existing Limitations

✓• Manual Team Formation – Friendships over skills, imbalanced teams.
✓• Limited Personalization – Faculty‑assigned topics, no skill alignment.
✓• Fragmented Communication – External tools cause inefficiency.
✓• No AI Support – No role assignment, task recommendation.
✓• No Industry Simulation – Students miss real‑world workflows.

✅ Proposed Solution

✓• AI‑Driven Team Formation – ML + NLP for balanced teams.
✓• Intelligent Task Assignment – Tasks aligned with strengths.
✓• AI‑Powered Project Generation – Custom ideas with milestones.
✓• Real‑Time Collaboration – WebSocket chat with presence.
✓• AI Mentor Chatbot – Contextual guidance and query resolution.

🎯 Aim & Objectives

🎯 Aim

✓• Automate team formation and task assignment.
✓• Provide AI‑based performance evaluation.
✓• Reduce the gap between theory and practice.

📋 Objectives

✓• Develop AI system for team formation and evaluation.
✓• Leverage ML and NLP for skill extraction.
✓• Implement chatbot for discussions and progress tracking.
✓• Simulate real‑world environments with gamification.

📚 Literature Survey

1. Silva et al. (2023) — Intelligent Tutoring System

Proposed Juh, a chatbot‑based tutoring system. Lacks team formation, task allocation, or end‑to‑end workflows.

Gap addressed: WorkExperio automates the full project lifecycle.

2. George (2024) — AI for Industry Readiness

Explores AI's role in bridging skill gaps; conceptual insights but no implementation.

Gap addressed: WorkExperio delivers concrete AI‑driven skill extraction and team balancing.

3. Bakhsh et al. (2024) — AI Collaboration for BA‑QA

AI‑assisted task coordination in professional settings, not academic collaboration.

Gap addressed: WorkExperio adapts AI tools for education with mentoring.

4. Pradeepa et al. (2023) — Resume Evaluation Tool

NLP for skill extraction; limited to pre‑project analysis.

Gap addressed: WorkExperio integrates resume parsing with team formation and project generation.

5. Ferrari & Muffatto (2024) — Generative AI for Startups

Genetic algorithms for startup team composition; not for academics.

Gap addressed: WorkExperio is tailored for academic settings with skill‑based clustering.

Summary: Existing systems address isolated components — tutoring, resume parsing, mentoring, or communication — but none offer an integrated, end‑to‑end solution for academic project execution. WorkExperio bridges this gap.

⚙️ Requirements Specification

🖥️ Hardware

✓• Dual‑core / Quad‑core CPU
✓• 4‑8 GB RAM (client)
✓• 2‑4 CPU cores, 8GB RAM (server)
✓• SSD storage
✓• Stable internet (≥5 Mbps)

💻 Software

✓• Python 3.11+ / FastAPI
✓• React.js / Tailwind CSS
✓• PostgreSQL / SQLite
✓• MongoDB
✓• Docker / Git
✓• VS Code / Postman

🔧 Functional

✓• User Authentication (JWT)
✓• Resume Parsing (PDF)
✓• AI Team Formation
✓• AI Project Generation
✓• Real‑Time Chat
✓• AI Mentor Chatbot

⚡ Non‑Functional

✓• Performance: <2s API response
✓• Security: JWT + bcrypt
✓• Scalability: Horizontal scaling
✓• Usability: Responsive UI
✓• Reliability: 99.9% uptime
✓• Maintainability: Modular code

🏗️ System Architecture

WorkExperio follows a multi‑tiered architecture across three primary layers:

🌐 Client Layer
⬇ React + Tailwind CSS + Vite ⬇
⚙️ Application Layer
FastAPI · WebSocket · AI Engine · JWT Auth
⬇ ⬇
🗄️ Data Layer
PostgreSQL (structured) + MongoDB (unstructured)

🧠 AI Engine Components

🧠 NLP & ML

✓• Resume Parsing
✓• Skill Extraction
✓• Team Formation
✓• Role Prediction

🤖 LLM & Automation

✓• Project Generation
✓• Task Assignment
✓• AI Mentor Chatbot
✓• Performance Analytics

🔌 Backend Services

✓• JWT Authentication
✓• Project Management
✓• Real‑Time Chat (WS)
✓• Task Tracking

📦 Implementation Modules (A to Z)

1. 🔐 Authentication Module

JWT + bcrypt for registration, login, and session management. Tokens stored in localStorage.

2. 📄 Resume Parser Module

Uses pdfplumber to extract skills, education, and experience from PDF resumes into structured JSON.

3. 🤖 AI Project Generator

LLM‑powered project idea generation with titles, descriptions, milestones, and required skills.

4. 👥 Team Formation

Clustering algorithms (K‑Means) for balanced, expertise‑based teams with role assignments.

5. 💬 Real‑Time Chat

WebSocket‑based instant messaging with typing indicators, online presence, and message history.

6. 🧠 AI Mentor Chatbot

Context‑aware conversational assistant for task explanations, guidance, and query resolution.

7. 📊 Project Management

Project creation, task assignment, progress tracking, and analytics dashboards with time logs.

8. 🗄️ Database Module

SQLAlchemy ORM with PostgreSQL (production) and SQLite (development) — 18 interconnected tables. Migrations via Alembic.

🧮 Algorithms & Pseudocode

🔐 Authentication (JWT + bcrypt)

Algorithm: Password Hashing & Token Generation
1. User submits email + password
2. Server hashes password using bcrypt
3. Stores hash in database
4. On login, compares provided password with stored hash
5. If valid, generates JWT (expires in 24h)
6. Returns token to client
7. Client stores token in localStorage
8. Subsequent requests include token in Authorization header

📄 Resume Parsing (NLP)

Algorithm: Skill Extraction from PDF
1. Upload PDF via POST /resumes/upload
2. Use pdfplumber to extract raw text
3. Clean text (remove special characters)
4. Apply NLP preprocessing (tokenization, stopword removal)
5. Match tokens against skill taxonomy
6. Extract education, experience, contact details
7. Return structured JSON
8. Store parsed_json in resumes table

👥 Team Formation (Clustering)

Algorithm: Skill-Based Team Clustering
1. Fetch all users with parsed skills
2. Convert skills to feature vectors (TF-IDF)
3. Define required skills for project
4. Compute similarity matrix (cosine)
5. Apply K-Means clustering
6. Assign roles based on top matching skills
7. Validate cluster balance
8. Return ranked team suggestions

🤖 Project Generation (LLM)

Algorithm: AI-Powered Project Idea Generation
1. Receive domain and skill inputs from user
2. Construct prompt for LLM
3. Send prompt to LLM API
4. Parse response for structured fields
5. Validate and clean output
6. Store generated project in database
7. Allow user to accept, modify, or regenerate
8. On acceptance, create project with milestones

🧪 System Testing

✅ Unit Testing

Test ID	Functionality	Result
UTC‑1	Password hashing during registration	✅ Pass
UTC‑2	Resume text extraction	✅ Pass
UTC‑3	AI project idea generation	✅ Pass
UTC‑4	Role assignment based on skills	✅ Pass

🔗 Integration Testing

Test ID	Integration Flow	Result
ITC‑1	Resume → Role Assignment	✅ Pass
ITC‑2	Team → Project Generation	✅ Pass
ITC‑3	WebSocket Chat → Storage	✅ Pass

📊 System Testing

Test ID	End‑to‑End Workflow	Result
STC‑1	Login → Upload → Team → Project	✅ Pass
STC‑2	Real‑Time Chat & Persistence	✅ Pass
STC‑3	AI Mentor Query & Response	✅ Pass

📊 Results & Performance

85%

Skill Extraction

90%

Role Prediction

250ms

API Response

100%

Message Delivery

The resume parser achieved 85% accuracy in skill extraction, and the role prediction model demonstrated 90% precision. Performance tests showed 250ms average API response, <200ms chat latency, and 100% message delivery reliability.

🚧 Challenges & Solutions

1. PDF Parsing Inconsistencies

Challenge: Different resume formats caused inconsistent extraction. Solution: Multi‑stage preprocessing with fallback OCR and regex normalization.

2. LLM Response Parsing

Challenge: AI‑generated projects sometimes missed required fields. Solution: Structured prompt engineering with JSON‑mode and auto‑retry with fallback.

3. WebSocket Connection Stability

Challenge: Frequent disconnections under high traffic. Solution: Heartbeat pings, auto‑reconnect with exponential backoff, Redis persistence.

4. Database Connection Hoisting

Challenge: ES6 import hoisting caused DNS resolution issues. Solution: Used require() and pool_pre_ping=True.

5. Skill Taxonomy Coverage

Challenge: Limited predefined skills caused missing extractions. Solution: Dynamic NER‑based skill extraction with manual user override.

🏆 Achievements

✓📄 Paper Accepted — SmartCom 2026 (Springer LNNS).
✓🏅 Hackathon — TechFrontier 2025, IIIT Delhi.
✓🧪 TRL 6 — Functional prototype validated in academic environment.
✓🌍 SDG 4 & 8 — Quality Education & Decent Work.
✓📊 85% extraction accuracy, 90% role precision, 250ms response time.

🔮 Future Enhancements

📈 Short‑Term (TRL 7)

✓• Personalized learning paths
✓• Emotion & sentiment analysis
✓• Advanced AI project mentor
✓• Predictive analytics

🚀 Long‑Term (TRL 8–9)

✓• Internship & recruiter matching
✓• Plagiarism detection
✓• Multi‑institution deployment
✓• Real‑time code collaboration

🔧 Technical

✓• Kubernetes orchestration
✓• Microservices architecture
✓• Real‑time code editor
✓• Video conferencing

📚 What I Learned

✓• Full‑Stack Development: React.js, FastAPI, WebSocket.
✓• AI & NLP: Resume parsing, skill extraction, clustering.
✓• Machine Learning: K‑Means for team formation.
✓• Database Design: PostgreSQL + MongoDB hybrid.
✓• LLM Integration: Prompt engineering and structured parsing.
✓• Performance Optimization: 250ms API response.
✓• Team Collaboration: Agile, Git, 4‑member team.
✓• Testing & QA: Unit, integration, system testing with 100% pass rate.
✓• Research & Publication: Paper writing and conference submission.

📝 Conclusion

WorkExperio successfully bridges the gap between academic learning and industry readiness by providing an AI‑driven, end‑to‑end project management platform. It integrates NLP‑based resume parsing, clustering‑based team formation, LLM‑powered project generation, real‑time communication, and an intelligent mentor assistant.

With 85% skill extraction accuracy, 90% role prediction precision, and 250ms average response time, the platform demonstrates both correctness and efficiency. The project has achieved TRL 6, with a paper accepted at SmartCom 2026 (Springer LNNS).