Dhurba Baral

• Fine-tuned BERT for document processing and implemented Neo4j to model data as a graph for representing and querying complex relationships between data entities.
• Fine-tuned Large Language Models (e.g., Llama and GPT) using Low-Rank Adaptation (LoRA) and quantization, and implemented Retrieval Augmented Generation (RAG) for building customer-facing chat bots.
• Implemented ML solutions using AWS SageMaker for end-to-end model training and deployment.
• Implemented CI/CD pipelines using Jenkins, writing Groovy scripts to automate build, test, and deployment workflows. Leveraged container-based technologies such as Docker to enable simplified, scalable, and reproducible production deployments.

Coursework: Regression, Classification, Clustering, and Arttificial Neural Networks, Recurrent Neural Networks, Transformers, and more.

Participated as Python and AI/ML Mentor in a month-long workshop on Machine Learning and Deep Learning forbachelor students.

Modules taught: Linear Regression, Logistic Regression, Naïve Bayes Classifier, Natural Language Processing (RNN, GRU, LSTM)

Built an AI-based system for detecting bees, monitoring their behavior and health status, and suggesting possible solutions to the detected problems directly to farmers.

Won the Best Hardware Project Award in the hackathon.

Tech Used: YOLOv5 for object detection module integrated with Raspberry Pi and Pi Camera, Convolutional Neural Networks (CNNs) for image classification

Participated as an ML instructor in a three-day workshop on Machine Learning and NLP.

Modules taught: Regression (Linear, Polynomial, Random Forest Regressor), Natural Language Processing (RNN, GRU, LSTM)

Participated in JENESYS 2016 program and studied cutting-edge technologies used in Japan for harnessing energy from various sources.