Harshit Makwana

Natural language processing / Case study

AI-Powered Interview Bot

I developed a two-round interview chatbot using TensorFlow, an LSTM-based seq2seq model, Flask and a custom HTML, CSS and JavaScript interface.

Harshit Makwana

Concept illustration of a conversation branching into aptitude and programming assessment paths.
AI-generated concept illustration · not a project screenshot
My role
Conversational model, Flask integration & frontend
Timeline
2022
Focus
Natural language processing

01 / Context

A conversational workflow for aptitude and role-specific assessment.

Interview assessment needs a structure that separates general aptitude from role-specific knowledge. I built a chatbot with two rounds: aptitude and reasoning first, followed by questions tailored to Python or Java job designations.

02 / Ownership

What I built

  • Implemented the conversational system with a seq2seq deep-learning approach in TensorFlow and an LSTM model.
  • Used historical and real-time data in the interviewing workflow.
  • Designed an aptitude/reasoning round and a second round tailored to the candidate's Python or Java designation.
  • Built the frontend in HTML, CSS and JavaScript and integrated it with a Python Flask backend.

03 / Engineering

Technical approach

Organize the assessment into two rounds

I separated the conversation into a general aptitude and reasoning stage and a job-specific stage. This gave the interview flow an explicit structure rather than treating every question as part of one undifferentiated conversation.

Connect the model to an application

I implemented the LSTM-based seq2seq model in TensorFlow and used Flask to integrate it with the interface. Building the frontend directly in HTML, CSS and JavaScript let me own the interaction layer alongside the backend and model integration.

04 / System view

Workflow at a glance

  1. Candidate interaction

    Custom web interface

  2. Conversation backend

    Flask integration

  3. Model & interview flow

    TensorFlow seq2seq with LSTM

  4. Assessment rounds

    Aptitude/reasoning, then Python or Java

05 / Delivery

Outcomes

I brought the model, backend and interface together into an interview chatbot with a defined two-round flow. The project demonstrates my ability to connect conversational modelling with a complete web application.

Technologies & methods

  • Python
  • TensorFlow
  • LSTM
  • Seq2seq
  • Flask
  • HTML
  • CSS
  • JavaScript

Automated Resume Insights

I built a Django-based resume parser using spaCy and NLTK to extract and organize skills, education, work experience and contact information.