Lochan Basyal

R&D: AI and Automation Special Project @ Walmart Inc.

IEEE/ACM Transactions on Audio, Speech, and Language Processing 2024 63rd IEEE Conference on Decision and Control (CDC) 2024 IEEE International Conference on Systems, Man, and Cybernetics (SMC) 2025 IEEE American Control Conference (ACC) IEEE SMC 2025 2026 IEEE American Control Conference (ACC)

Train the next generation of AI with RLHF

Lead the creation of advanced object detection and segmentation models, fine-tuning YOLO pre-trained models with diverse Roboflow datasets for an exceptional 97% accuracy. Worked on the multimodal, developing a visual question-answering system utilizing Hugging Face and Streamlit. Drove the development of intricate crowd-counting models, collaborating across teams to seamlessly integrate cutting-edge computer vision solutions. Achieved a remarkable 95% accuracy rate, solidifying project success. Owned the complete MLOps lifecycle, including data monitoring, code refactoring, and the development of robust model monitoring workflows for efficient model lifecycle management. Collaborated with cross-functional teams to define AI project requirements and objectives, ensuring alignment with overall business goals.

Gen AI, LLMs Research

Developed an end-to-end machine learning solution on GCP Vertex AI, encompassing data ingestion, model training, validation, testing, and deployment, fostering collaboration across teams, and ensuring scalability, security, and continuous improvement Leveraged LangChain and OpenAI to extract insights from PDF documents, using LangChain’s PyPDFLoader for efficient page handling, and applied generative AI for succinct text summarization, streamlining information extraction and enhancing accessibility Scraped over 2000 Tweets from Twitter API to fetch the sentiment type using NLTK’s Sentiment Intensity Analyzer and Text Blob, pre-processed the text to eliminate unwanted elements from tweets for extracting the new features and generate the word cloud to visualize the common words in positive and negative tweets Conducted comprehensive data pre-processing and visualization on wine review datasets, revealing invaluable insights into the underlying patterns and trends. Implemented a powerful multi-label classification model leveraging a BERT pre-trained model, enabling accurate classification of diverse wine varieties and achieved an impressive 84% overall accuracy with hyperparameter tuning Analyzed the PatchCamelyon benchmark datasets to train in the different models, multilayer perceptron, convolution, ResNet50, InceptionV3, and ensemble to achieve the model’s performance with 95% accuracy Worked with big data technologies, SQL, and data science tools

Developed and implemented generative AI models, leveraging deep learning techniques such as GPT, VAE, and GANs, to achieve specific objectives Owned the complete MLOps lifecycle, including data monitoring, code refactoring, and the development of a robust model monitoring workflows for efficient model lifecycle management Designed and implemented scalable end-to-end data science solutions for data products, collaborating closely with data engineers and data analysts Demonstrated strong thought leadership by advising product and business stakeholders on the development, scaling, and deployment of machine learning solutions Implemented a machine learning model to optimize the company’s recommendation system, resulting in a 20% increase in customer feedback, identifying key patterns and trends. This initiative led to a 30% reduction in customer churn rate within six months Collaborated with cross-functional teams to design and deploy a fraud detection system, utilizing anomaly detection algorithms. This solution reduced fraudulent activities by 40%, resulting in substantial cost savings for the company Conducted extensive data analysis on user behavior and preferences, which informed the redesign of the company’s mobile application. As a result, user retention increased by 35%, and app ratings improved from 3.5 to 4.5 stars

Concentration: Artificial Intelligence (AI)