- Designed an MLOps framework on GCP, streamlining the ML lifecycle from development to deployment, scaling, and maintenance.- Deployed production-ready ML models using Docker and Kubernetes, ensuring high availability and scalability.- Automated model development, tuning, and serving with Vertex AI, improving efficiency.- Utilized Langsmith and Langchain to deploy LLMs, generate synthetic datasets, fine-tune models, and evaluate performance.- Achieved a 15% accuracy improvement with state-of-the-art ML architecture.- Reached 90% code coverage with unit, integration, and performance testing.- Reduced deployment time by 30% through CI/CD pipeline automation using Terraform.

- Research and Development to automate manual framework mapping tasks using transformers and BERT architecture.- Built ETL pipelines for document classification.- Research and Development for using natural language processing techniques to convert unstructured cyber security documents (PDF, doc, PPT) to structured documents.

- Created modules for image classification, object detection, image de-duplication, and image difference detection.- Collaborated on developing Natural Language Processing (NLP) algorithms for PIN.- Developed DevOps pipeline to create an application package of image cognition modules using Docker.- Evaluated AWS services including IAM, EC2, EKS, Fargate, and Sagemaker.

- Appian developer and designer of the development team working for a bank in UAE.- Implemented best practices and maintained code quality across the projects.- Worked with Relational Databases to provide backend support.- Proficiently used agile project management tools such as Jira.

- Worked on pipelines to automate the process of data mining and modelling.- Created a feedback sentiment analysis model, including sentiment analysis of entities, using libraries such as nltk, spacy.- Actively involved in creating a seq2seq conversation model using hcn network. Was responsible for creating dynamic word vectors for training data. - Integrated voice with the seq2seq conversation model to carry out the conversation between bot and human entirely through voice.- Worked on solving the problems of coreference and handling of complex sentences during chatbot conversations.