NLP
TrendED Pathfinder: career recommender
NLP role matching and skill-gap learning paths
- Timeline
- Sep 2024 to Nov 2024
- Affiliation
- Worcester Polytechnic Institute
- Team
- Sakshi, Blake, Harrison, Ankit
- Job descriptions
- 10,000+
- Match accuracy
- 95%
- NLP precision
- 90%
What it is
An NLP engine that matches resumes to roles with TF-IDF and cosine similarity, calibrates rankings with supervised classifiers, and recommends courses that close the skill gaps it finds.
Problem
Job boards filter on job type and company, not on what a specific student can do and should learn next.
Approach
Scraped job boards and university catalogs with BeautifulSoup, built the matching and ranking models in TensorFlow and scikit-learn, and generated top-k role plus coursework recommendations from unified text features.
Highlights
Scaled to 10,000+ job descriptions, 5,000 resumes and 300+ course catalogs.
95% match accuracy with 90% NLP precision; irrelevant results down 50% and data preparation time down 65%.
Used by 1,000+ students for tailored course suggestions; LDA topic modeling on job and catalog text.