🤖 Corporate Companion Chatbot

A LangChain-based LLM-powered chatbot to assist employees with internal organizational tasks including personal data feeding and querying, meeting scheduling, and intelligent file organization and other core functionalities.

• Collects:
  • Employee ID (Unique identifier for employees)
  • Name
  • Contact Details (Email, Phone)
  • Optional fields: Department, Office Location
  • Resume (PDF)
• Handles missing data gracefully by storing placeholder values.
• Validates critical details like emails and phone numbers
• Parsing of phone numbers (India format) and emails into proper structure for storage purposes.

• Schedules meetings:
  • With individual employees
  • With multiple employees
  • With entire teams
• Checks availability using data files:
  • employee_schedules.csv (booked slots)
  • employee_teams.csv (team membership)
• Considers the following aspects:
  • Office hours (9 AM – 6 PM, Mon–Fri) should be considered for meetings
  • Lunch breaks (1 PM – 3 PM) should not be considered
  • 1-hour unavailability rule i.e. if the booked slot is for 10:00 AM, then the person will be unavailable for the next 1 hour
• Finds earliest valid time slot for all participants.

• Uses an LLM to:
  • Classify available files into categories (e.g., Finance, HR)
  • Create category folders
  • Move files accordingly
• Categorizes based on the name of the file (can be extended to checking of file content as well).

• Answers the queries of the users related to the HR policies and holidays
• Showing up of upcoming events on being asked by the user
• The data like policies, holidays information, events, etc. is hardcoded in the file (later on, can be extended to the functionality of being fetched from the database) -In production level, the data will be stored somewhere else and will be fetched into the LLM for answering purposes.

• Python
• LangChain
• Streamlit (UI)
• Pydantic (structured output parsing)
• transformers / HuggingFace (LLM backend)
• dotenv (for environment variable handling)
• Pandas, os, shutil, json (file & data handling, data loading)
• typing (for type annotations)
• datetime (for date-time handling and manipulations)
• re, phonenumbers, email_validator (for validation purposes)

Follow the steps below to set up and run the Corporate Companion Chatbot on your local machine.

git clone https://github.com/ashutosh229/corporate-companion.git
cd corporate-companion

python3 -m venv venv
source venv/bin/activate

pip install -r requirements.txt

• Create a .env file in the root of your directory
• Load the following env variables in the .env file

REPO_ID = "google/flan-t5-small"
TASK = "text2text-generation"
TEAMS_FILE = "data/employee_teams.csv"
SCHEDULE_FILE = "data/employee_schedules.csv"
SAMPLE_FILES_DIR = "data/sample_files"
FILE_CATEGORIES_DIR = "data/categories"
HUGGING_FACE_TOKEN = "your-huggingface-token"
USER_DATA_DIR = "data/user_data"

cd data
mkdir categories sample_files user_data
cd user_data 
mkdir resumes
cd .. 
cd ..

• Create 10 sample files in the 2 logical categories: Finance and HR in the /data/sample_files directory

streamlit run app.py

A detailed journal has been maintained for this project. It includes:

• The difficulties faced during the development of the project
• What worked and what did not
• What could be improved in the implementation
• Pending implementations

📄 Access the journal here: View Journal on Google Docs

The solution for the problem 2 of the assignment includes the following:

• Documentation for the Python request library
• Documentation for scraping flat listings from Magicbricks for the Delhi region

📄 Access the solution here: View solution on Google Docs

If you have any questions, suggestions, or feedback about this project, feel free to reach out:

Name: Ashutosh Kumar Jha
Email: Email
LinkedIn: Linkedin
GitHub: Github

⭐ If you found this project helpful, feel free to give it a star on GitHub!