BNVS Kumar

Improved and maintained a robust JD parser that required converting unstructured data into structured data and extracting useful information using techniques of NLP. Trained BiLSTM CRF based NERs with character embeddings and Attention Mechanism delivering high recall and precision (>92% F1) for recognizing various entities such as location, degree, client-facing, job title, etc. Extracted relevant keywords from the job description, and recommended them on the resume to enhance its alignment with specific job requirements. Utilized GPT-2 to generate sample bullets for recommended keywords and deployed the model using TorchServe in SageMaker for efficient model serving. Enhanced the user’s understanding of job requirements by utilizing data extraction and sanitization techniques to identify and incorporate key requirement elements such as job title, job level, and required skills from the job description. Utilized containerization techniques to deploy microservices on AWS, managing the containers with AWS Copilot, and integrating a CI/CD pipeline to automate the software development life cycle, thereby enabling faster Time-to-Market. Performed load testing using Locust to evaluate system performance, identify bottlenecks and optimize resource allocation for enhanced scalability. Leveraged AWS SNS/SQS services to enable asynchronous communication between microservices, thereby enhancing the system performance.

Analyzed various deep learning models using transformers, seq2seq network with attention mechanism for generating questions from a news article. Evaluated encoder-decoder models for abstractive text summarization task and used ALBERT (A Lite BERT) for question answering task.