How thanushamh/talentscout-chatbot works
TalentScout is a Streamlit hiring-screening chatbot. A candidate enters a chat, is guided through profile collection and validation, then receives technical questions generated from their role, experience, and tech stack. The app keeps candidate data and chat history in session state. Its LLM utility supports local TinyLLaMA through Ollama and a mock-question mode; the README also describes a question-only display after profile submission. The graph follows visible calls and documented behavior; it does not assume additional storage or service integrations.
Main entry points: app.py Screening flow · utils.py LLM response selector · prompts.py Prompt definitions
Components 9
Candidate conversation
- Screening flow
app.py - Profile collection
app.py - Session state
app.py
Question generation
- Prompt definitions
prompts.py - LLM response selector
utils.py - TinyLLaMA via Ollama
utils.py - Mock questions
utils.py - Technical questions
app.py
Other components
- Candidate
Connections 11
- Candidate → Screening flow: enters answers
app.py - Screening flow → Profile collection: routes answers
app.py - Profile collection → Session state: stores details
app.py - Screening flow → Session state: stores chat
app.py - Screening flow → Prompt definitions: imports prompts
app.py - Screening flow → LLM response selector: requests questions
app.py - LLM response selector → TinyLLaMA via Ollama: selects local model
utils.py - LLM response selector → Mock questions: selects mock mode
utils.py - TinyLLaMA via Ollama → Technical questions: returns output
utils.py - Mock questions → Technical questions: returns questions
utils.py - Screening flow → Technical questions: displays output
app.py