Rhitesh Kumar Singh

Background

About Rhitesh Kumar Singh

As an NLP enthusiast and machine learning practitioner, I am passionate about exploring and implementing cutting-edge technologies in natural language processing and machine learning. With a strong foundation in both classical and modern methods, I have developed custom NLP models for various applications, particularly in question answering- I developed a QA chatbot using RAG with Langchain and evaluated RAG pipelines using the Ragas framework. Additionally, I have successfully built a seq2seq BiLSTM model with self-attention, cross-attention, and used BERT for the QA task on the SQuAD dataset- In addition to my NLP work, I have hands-on experience in building classical machine learning models from scratch. My projects include developing decision trees, PCA, Gaussian Mixture Models, and linear regression models. I have applied these models in various applications, such as Image Captioning using KNN and FAISS, Background Subtraction with Gaussian Mixture Models (GMM), and Face Recognition and Reconstruction with PCA- I have also explored transfer learning, fine-tuning models like InceptionV3, Vision Transformers, and DeiT for various tasks. In a private Kaggle competition on age prediction from facial images, I placed in the top 3% by using transfer learning on ViT and DeiT, achieving a winning MAE of 5.15 on the public test set and 4.90 on the private test set, securing a top 5 finish- I am currently working on LLM security, Red Teaming LLMs, and Responsible AI- Explore my projects and contributions on GitHub: https://github.com/RhiteshKS