Pascal Rai

Experience

• 

  Code Himalaya

  Dec 2022 - Apr 2024

  • Explored and implemented various CV CNN architectures, tracing their evolution from ImageNet, VGG, and ResNet to Inception, MobileNet, EfficientNet, and ViT, ensuring efficient model selection for diverse tasks.• Mentored interns and juniors on computer vision, chatbot development using Rasa and open-source generative language models, and guided them in building production-ready models.• Developed a module for integrating an Elasticsearch vector database into a RAG-based chatbot, enhancing its semantic search capabilities.• Computed and visualized stock metrics using pandas, date-time manipulation, and Elasticsearch's Python client, indexing time series data and visualizing the metrics via Streamlit.• Created a custom dataset and trained a Siamese network using ResNet with triplet-loss, achieving an f1 score of 0.989 for comparing similar and dissimilar images of varying resolution.• Built a pipeline in Python for a large-scale automatic mailing service triggered by a cron job, filtering return emails and extracting entities using a comparison of Spacy and custom BERT NER for entity extraction. Show less • Implemented ML libraries like Pandas for Data Analysis, Matplotlib for data visualization, and TensorFlow for Deep Learning models.• Familiarized with various Loss functions, Optimization functions, Evaluation metrics, and Hyper-parameter tuning.• Augmented images and used VGG16 transfer learning to extract features from images to train a Custom object detection and keypoint detection model.• Made a Face Recognition model using OpenCV DNN model to extract faces, Facenet pre-trained model to extract face embeddings. Show less

  • Junior AI Developer

    Apr 2023 - Apr 2024

  • AI Developer Trainee

    Feb 2023 - Mar 2023

  • AI Developer Intern

    Dec 2022 - Jan 2023
• 

  Frost Digital Ventures

  Apr 2024 - now

  Associate AI Engineer

  • Engineered intelligent workflows with LLM technologies, FastAPI, and containerization, achieving about 95% reduction in resource usage.• Deployed advanced embedding models, including a semantic recommendation engine, enhancing scalability and efficiency.• Developed and deployed custom scalable Detection Models as a GPU Containerized service.• Streamlined and optimized AWS cloud services, cutting costs by 50% while preserving service quality, scalability, and reliability.• Managed Kubernetes deployments and server migration to local cloud providers, improving system performance and stability. Show less

• 

  Freelance

  May 2024 - Jun 2024

  Cloud DevOps Architect

  • Containerized deployment of a 3-tier .NET application with MSSQL database.• Implemented automated CI/CD pipelines using Jenkins and Docker.• Secured domain with SSL certificates and configured Nginx for reverse proxy, enhancing security and traffic management.• Optimized AWS services for cost-efficiency and performance, and ensured periodic database backups.• Managed DNS records for seamless server connectivity and set up domain validation for domain-based email communications.