Cygnus

Inspiration Finding the right job in India can be overwhelming. Many platforms list thousands of openings, but they don’t tailor them to candidates’ unique skills, goals, or experiences. We were motivated by the challenges fresh graduates and young professionals face, like customizing resumes, identifying skill gaps, and navigating endless job portals. This inspired us to create Cygnus, an AI-powered career assistant. It simplifies resume improvement, skill analysis, and job discovery while offering useful career insights.

What We Learned Agentic AI workflows can coordinate several independent agents to achieve complete automation. Resume parsing and standardization are crucial for fair and accurate job matching. Tools like Selenium can scrape job data from career portals, allowing for real-time opportunities. IBM Granite models drive contextual embeddings and semantic matching to link candidates with roles beyond just keyword searches.

How We Built It Resume Parser Agent: Extracts structured data like skills, education, and experience using Granite NLP. Skill Gap Analyzer: Compares against job descriptions and suggests upskilling paths. Resume Optimizer: Creates personalized resume suggestions. Job Discovery Agent: Uses Selenium to explore multiple job portals and gather results. Workflow Management: Handled with the IBM Agent Development Kit (ADK). UI Layer: Offers an interactive dashboard for candidates to see jobs, resume scores, and suggested skills.

The workflow follows a complete automation pipeline: Resume Input → AI Parsing → Skill Analysis → Resume Optimization → Job Matching

Challenges We Faced Resume Diversity: Formatting differences made parsing difficult. Job Data Availability: Not all portals provide APIs; we automated scraping with Selenium. Contextual Matching: Mapping skills to job roles required fine-tuning embeddings. Workflow Management: Coordinating multiple AI agents with ADK while maintaining consistent performance benchmarks.

Built With

• adk
• dataset-cleaning-selenium-?-automated-job-portal-crawling-performance-benchmarks-?-for-agent-efficiency
• embeddings
• granite
• javascript
• javascript-frameworks/libraries:-pytorch
• postgre
• python
• pytorch
• resume-parsing-ibm-adk-?-agent-workflow-orchestration-ibm-data-preprocessing-toolkit-?-resume-standardization
• scikit-learn
• selenium
• selenium-cloud-services:-ibm-cloud-(granite-models
• skill-gap-accuracy
• toolkits)-databases:-postgresql-for-structured-candidate/job-data-technologies-&-apis:-ibm-granite-models-?-nlp