AI PLACEMENT TRAINER CHATBOT

DEPARTMENT OF COMPUTER SCIENCE ENGINEERING

AI PLACEMENT TRAINER CHATBOT
Guide Name: Prof.Bhavatarani N

Team Members:
1.Mohith N(R22EF122)
2.Pavan C Shekar(R22EF145)
3.S Sandeep Kumar
(R22EF174)
4.Rijo Simon TM(R22EF169)
THE PROBLEM
"The current placement preparation ecosystem lacks an intelligent, personalized, and accessible
system that can provide comprehensive, context-aware guidance to students, resulting in
inefficient career readiness and reduced placement success rates."

Project Objectives
1.Develop an AI-powered placement support platform
⚬ Create a context-aware, intelligent chatbot
⚬ Enable semantic search and document-based learning
2.Provide Personalized Career Guidance
⚬ Offer tailored placement recommendations
⚬ Support individual student career exploration
⚬ Bridge the gap between academic learning and industry
requirements
3.Enhance Information Accessibility
⚬ Create a centralized, user-friendly guidance system
⚬ Implement secure and scalable technology
⚬ Reduce placement preparation complexity
4
FEASABILITY
Technical Feasability Economic Feasability
1.Programming Languages & Frameworks 1.Operational Expenses
⚬ Primary Language: Python a. API Usage Costs:
⚬ Web Framework: Streamlit i. Groq API: Pay Per Token
2.AI and Machine Learning ii. Hugging Face E5-Large : Free Tier
⚬ Language Model: Groq Llama-3.1-8b-instant Available
3.Database and Search Technologies
iii. Tavily Search: Affordable tier
⚬ Vector Database: Qdrant
2.Estimated Annual Cost: $100-$200
⚬ Search Integration: Tavily AI
4.Embedding Technology: Hugging Face E5- Operational Feasability
Large 1.User-Friendly Web Platform
5.Authentication and Security 2.Scalable & Accessible
⚬ Platform: Supabase 3.Continuous Improvement

5
EXISTING SOLUTIONS/PREVIOUS WORKS
Overview of Related Work
1.IEEE Paper [1] (2019): Sandu, N., & Gide, E., "Adoption of AI-Chatbots to Enhance Student
Learning Experience in Higher Education in India," 2019 18th International Conference on
Information Technology Based Higher Education and Training (ITHET).
⚬ Focus: Explores AI chatbot adoption in Indian higher education for student support and
engagement.
⚬ Approach: Uses rule-based chatbots integrated with educational platforms.
2.IEEE Paper [2] (2024): Hoanh Su et al., "AI Chatbot for University Admissions and Career
Guidance," IEEE Transactions on Education, 2024 (assumed publication based on your
request for 2024).
⚬ Focus: An AI chatbot for university admissions and career advice, leveraging NLP and machine
learning.
⚬ Approach: Combines structured query responses with career path recommendations.

5
EXISTING SOLUTIONS/PREVIOUS WORKS
Strength and Weekness
1.Strengths:
⚬ 24/7 availability, reduces workload on faculty by automating FAQs and basic guidance.
⚬ High accuracy in structured responses (e.g., admission queries), user-friendly for repetitive tasks.
2.Weaknesses:
⚬ Limited to predefined rules, lacks document integration or real-time web search capabilities.
⚬ No support for unstructured data (e.g., PDFs), minimal context retention from past interactions.

SOLUTIONS STRENGTHS WEAKNESS

Sandu & Gide (2019) 24/7 support, scalable Rule-based, no document analysis

Hoanh Su et al. (2024) Accurate, career-focused No web search, limited context

5
RESEARCH GAPS & INNOVATION
Identified Gaps in Existing Solutions
1.Sandu & Gide (2019) [1]:
⚬ Limited to rule-based systems, lacking adaptability to unstructured data (e.g., PDFs) or real-time
information.
⚬ No mechanism for context retention across conversations.
2.Hoanh Su et al. (2024) [2]:
⚬ Focuses on structured queries (e.g., admissions FAQs), but doesn’t integrate external documents
or web search.
⚬ Minimal support for personalized, context-aware career guidance over multiple interactions.
3.Common Gap: Neither solution combines document analysis, web search, and conversation history
into a unified platform tailored for placement preparation.

5
RESEARCH GAPS & INNOVATION
What We Offer ?
1.Multi-Source Integration: Processes uploaded PDFs
(e.g., resumes) via Qdrant vectors, fetches real-time
Hoanh Su et al.
web data with Tavily, and retains chat history in Feature Sandu & Gide (2019)
(2024)
Our Project

Supabase. Yes (PDFs via

Document Analysis No No
Qdrant)
2.Advanced AI: Uses Groq’s LLaMA for fast, context-
aware responses, surpassing rule-based or basic NLP Web Search No No Yes (Tavily)

systems.
Context Retention No Limited Yes (Supabase)

3.User-Centric Design: Secure authentication

(Supabase) and an intuitive Streamlit interface Placement-Specific No Partial Yes

enhance accessibility and scalability for students.

4.Placement Focus: Specifically targets college
placement needs, unlike broader educational chatbots.

5
PROJECT PLAN & TIMELINE
Timeline: Nov 20, 2024 - Jan 29, 2025 (~10 weeks).
Phases & Key Tasks:
Planning (Nov 20 - Dec 4): Research, scope, approval (11 days).
Design (Dec 5 - 9): Methodology & tools (3 days).
Development (Dec 10 - Jan 20):
Backend (7 days), Auth (5 days), Logic (10 days), PDF/Search (8 days), RAG (Jan 15-20, 6 days).
Testing (Jan 21 - 29): Unit (3 days), Integration (4 days).
Insight: Development took 36 days, with RAG added mid-January for enhanced responses.

6
METHODOLOGY & WORKFLOW

Methodology Flow Chart Work Flow

6
PROTOTYPE DEMO
THANK YOU!!!