-
-
Bot in action
Career Coach AI
💼 Your personal AI career advisor — built with FastAPI, powered by Amazon Bedrock Agents.
🌟 Inspiration
During my own career transitions, I realized how difficult it is to get personalized, human-like career guidance.
Most bots give generic advice — none truly understand context, tone, or emotional intent.
I wanted to build a real AI career coach capable of:
- Rewriting resumes and LinkedIn summaries
- Generating role-specific cover letters
- Offering personalized career insights
- Scheduling mock interviews (future goal) All from a simple Telegram chat, powered by Amazon Bedrock Agents.
💬 What It Does
Career Coach is an AI-powered Telegram bot that acts as a virtual career advisor.
It helps users:
- ✍️ Improve resumes and optimize job applications
- 💼 Write personalized cover letters
- 💬 Practice interview questions and soft-skill responses
- 📅 (Planned) Schedule interview prep sessions automatically
Each message is processed through Amazon Bedrock, where an Agent:
- Chooses which model to use (Claude 3 or Titan)
- Decides what reasoning steps to take
- Produces an actionable and empathetic response
🧱 How We Built It
🧩 Tech Stack Overview
| Component | Description |
|---|---|
| FastAPI | Serves Telegram Webhook + Bedrock requests |
| Amazon Bedrock | Runs Claude 3 / Titan models and Agents |
| Telegram Bot API | Provides user interface |
| Railway | Simplifies deployment and environment configuration |
| python-telegram-bot v20 | Async message handling |
| .env | Secure secret storage |
🧮 System Architecture Flow
$$ \text{User Message} \xrightarrow{\text{Telegram Webhook}} \text{FastAPI Server} \xrightarrow{\text{AWS Bedrock Agent}} \text{Claude 3 / Titan} \xrightarrow{} \text{AI Reply to User} $$
🧰 Built With
Python, FastAPI, Telegram Bot API, Amazon Bedrock (Claude 3 Haiku + Titan Text Lite),
Uvicorn, Railway,GitHub
🚧 Challenges We Ran Into
- Handling async initialization for the Telegram bot lifecycle
- Managing AWS Bedrock tokens securely on Railway
- Designing prompts for consistent tone and quality
- Balancing inference speed and response accuracy
- Debugging async chains between FastAPI, Bedrock, and Telegram
🏆 Accomplishments We’re Proud Of
- 🚀 Fully functional, production-ready Telegram bot
- 🤖 Real integration with Amazon Bedrock Agents
- ⚙️ Clean modular architecture with reusable components
- 🧠 Context-aware AI conversations with empathy and precision
- 🌐 Live deployment on Railway with near-zero downtime
📚 What We Learned
- How to orchestrate multi-step reasoning with Amazon Bedrock Agents
- Prompt engineering for Claude 3 vs Titan
- Best async practices with FastAPI + Telegram Bot API
- Modular backend design for AI-driven chat systems
- Delivering technical depth with human-centered UX
🚀 What’s Next for Career Coach
- 🧠 Add session memory and user context for personalization
- 💼 Integrate job-search APIs and Google Calendar scheduling
- 🌍 Expand multilingual support (English, Russian, Spanish, Hindi)
- 📊 Add analytics dashboard for usage insights
- 🪄 Inline Telegram buttons for “Rewrite / Summarize / Improve” options
🧮 Future Math Feature Example
If we later introduce a resume scoring system, we can model it as:
$$ \text{CareerScore} = w_1 \cdot \text{Experience} + w_2 \cdot \text{Skills} + w_3 \cdot \text{Education} $$
where
$w_1, w_2, w_3$ are learned weights representing each category’s importance.
This can evolve into a data-driven evaluation engine for ranking resumes or candidates.
🧑💻 Hackathon Demo Script
🎤 Elevator Pitch
Career Advisor AI — a Telegram bot that helps you improve your resume, write cover letters, and prepare for interviews — powered by Amazon Bedrock Agents.
🧩 Demo Flow
Open Telegram → Start the Bot
Type/start— the bot introduces itself.Ask a Career Question
👩💻 User: Write me a cover letter for a Product Manager position at Google. 🤖 Bot: Dear Hiring Manager, I’m excited to apply for the Product Manager role at Google...Agent in Action
- Bot sends your prompt to Bedrock Agent
- Agent chooses Claude or Titan based on complexity
- Generates a structured response
Show Live Logs
- Display how FastAPI and Bedrock integrate in real time
End Note
- “Built with FastAPI + Telegram + Amazon Bedrock Agents on Railway”
📬 Contact
Made with ❤️ by @evgeniyabautina
Contributions, ideas, and stars ⭐ are always welcome!
Log in or sign up for Devpost to join the conversation.