Inspiration
Applying for jobs today is more complicated than ever. Candidates must tailor resumes for different roles, understand job requirements, identify missing skills, generate professional outreach messages, and keep track of dozens of applications. Many students and early-career professionals struggle with this process. We wanted to build an AI-powered career assistant that acts as more than a chatbot—it actively helps users make better career decisions and manage their job search journey efficiently.
What it does
AI Career Agent helps users throughout the job application process by:
- Analyzing resumes against job descriptions
- Calculating ATS-style match scores
- Identifying missing skills and knowledge gaps
- Generating personalized learning roadmaps
- Creating professional referral and networking messages
- Tracking job applications and statuses
- Storing application history and analysis results in MongoDB
- Executing multi-step workflows using AI reasoning
The platform transforms career guidance into an intelligent, action-oriented experience.
How we built it
We built AI Career Agent using a modern AI application architecture:
- Frontend: Streamlit for an interactive user experience
- Backend: FastAPI for scalable API services
- AI Layer: Gemini through Google Cloud Vertex AI with local fallback support
- Database: MongoDB for persistent storage of resumes, analyses, applications, and agent traces
- Agent Logic: Multi-step workflow orchestration that performs analysis, recommendations, message generation, and data persistence
The agent parses resumes, analyzes job descriptions, computes match scores, generates recommendations, and stores results in MongoDB for future interactions.
Challenges we ran into
One of the biggest challenges was building an agent that performs meaningful actions rather than simply generating responses. We needed the system to execute multiple steps, maintain context, and store information for future use.
Another challenge was integrating multiple components, including resume parsing, job analysis, referral generation, and MongoDB persistence, into a seamless workflow. Ensuring the application remained functional even without cloud credentials also required implementing a reliable local AI fallback mechanism.
Accomplishments that we're proud of
- Built a complete end-to-end AI career assistant
- Successfully integrated MongoDB as a persistent memory layer
- Implemented multi-step agent workflows instead of simple chatbot interactions
- Developed ATS-style resume and job matching capabilities
- Added automated referral message generation
- Created a scalable architecture that can be extended with additional career services
- Delivered a production-ready prototype within the hackathon timeline
What we learned
During development, we gained valuable experience in:
- Building AI agents capable of reasoning and task execution
- Integrating Gemini and Google Cloud services
- Designing scalable FastAPI applications
- Using MongoDB for long-term memory and workflow persistence
- Combining traditional NLP methods with modern LLM capabilities
- Creating practical AI solutions that solve real-world problems
We also learned how important structured data and persistent storage are for making AI agents truly useful.
What's next for AI Career Agent
Our vision is to evolve AI Career Agent into a complete AI-powered career platform.
Future enhancements include:
- Real-time job discovery and recommendations
- Automated application management
- Interview preparation and mock interview agents
- Personalized career coaching
- Skill development tracking and certification recommendations
- Multi-agent collaboration for resume optimization and career planning
- Integration with professional networking platforms
Ultimately, we aim to create an intelligent career companion that supports users from skill development to securing their dream job.
Built With
- agentic-ai
- ats-matching-engine
- career
- css
- fastapi
- gemini-api
- git
- github
- google-cloud-agent-builder
- google-cloud-platform-(gcp)
- google-cloud-vertex-ai
- google-gemini
- html
- javascript
- mongodb-atlas
- mongodb-atlas-api
- mongodb-mcp-server
- multi-step-agent-workflow
- natural-language-processing-(nlp)
- numpy
- pandas
- pymongo
- python
- python-dotenv
- recommendation
- requests
- resume-parsing
- scikit-learn
- streamlit
- tf-idf
- uvicorn
- vertex-ai-api
- vs-code
Log in or sign up for Devpost to join the conversation.