I am facing a challenge in matching 2 million resumes to 200 active job openings with a target accuracy of 80%. I aim to streamline our recruitment process and ensure accurate candidate-job matching. Here are the details:

Problem Statement
Objective: Efficiently match a large number of resumes to a smaller number of job descriptions.
Goals:
High Accuracy: Target at least 80% accuracy.
Scalability: Handle large volumes of data.
Automation: Minimize manual intervention.
Possible Technologies and Approaches
Natural Language Processing (NLP):
Text Preprocessing: Cleaning, tokenization, and normalization.
Feature Extraction: TF-IDF, Word2Vec, GloVe, BERT embeddings.
Machine Learning Models:
Supervised Learning: Training on labeled data.
Unsupervised Learning: Clustering techniques.
Deep Learning Models:
Transformer-based Models: BERT, RoBERTa for better contextual understanding.
Similarity Measures:
Cosine Similarity.
Advanced Metrics: Custom metrics tailored to data.
Tools Used
Vector database
OpenAI’s Ada-002 model for embeddings
Request for Expertise
Has anyone worked on a similar problem or used the mentioned technologies? Any insights or advice on the best approach to achieve our accuracy target would be greatly appreciated.

Thank you!