Santosh T Email and Phone Number

Implemented a novel RAG architecture for a company chatbot leveraging GenAI's pre-trained LLMs and retrieval capabilities to ensure retrieval of the most relevant documents for each user interaction. Utilized Llamaindex capabilities for query expansion and synonym matching, enhancing the chatbot's ability to interpret user intent and retrieve relevant information.Worked on embedding models by selecting and utilizing appropriate techniques (e.g., Sentence Transformers, Universal Sentence Encoder) to create high-quality vector representations of the company's documents.Experimented with variations in prompt formulations within GenAI framework to assess their impact on response diversity and user engagement, aiming for a more natural and varied conversation flow.Successfully implemented a vector database (Pinecone) to store and manage document embeddings, enabling scalable and fast nearest-neighbor searches at response time.Optimized vector database configuration and query strategies to align with the specific retrieval needs of the chatbot, ensuring optimal performance and user experience.Ensured accurate semantic similarity comparisons for efficient retrieval of relevant information within the RAG framework, even when dealing with varied document lengths and complexities.Applied Prompt Engineering techniques to fine-tune Large Language Models (LLMs), enhancing the chatbot's ability to generate contextually relevant and coherent responses.Adopted an iterative approach to fine-tuning, incorporating feedback from model evaluations and continuously refining the model architecture for improved results.Leveraged Streamlit's capabilities for rapid prototyping and iterative development, accelerating the design and testing of new chatbot features.

Utilized SageMaker pre-built tools and pre-trained models to accelerate development and experimentation, reducing overall time-to-market for ML solutions.Successfully deployed ML models on SageMaker for web services, batch jobs, and real-time inference, ensuring scalability and responsiveness in production environments.Implemented routing using Python frameworks such as Flask for request routing and dispatching, efficiently handling incoming requests and directing them to their respective handlers.Developed custom libraries and utilized standard Python packages to promote best practices in software development, contributing to the enhancement of coding standards.Experience in python, Jupiter, Scientific computing stack (NumPy, SciPy, pandas and matplotlib).Generating various capacity planning reports (graphical) using Python packages like NumPy, matplotlib.Experience in using Scikit-Learn and Stats models in Python for Machine Learning and Data Mining.Created custom pipelines for feature engineering, data preprocessing, and model evaluation using scikit-learn.Applied advanced scikit-learn techniques like hyperparameter tuning or ensemble methods to optimize model performance and generalizability.Developed microservices in Python to process data from various databases, including PostgreSQL and Datastore, showcasing expertise in Python-based microservices architecture.Designed and implemented custom dashboards using Python visualization libraries for Grafana to monitor and visualize Prometheus metrics effectively.

Involved in all aspects of the Software Development Lifecycle process (SDLC): gathering requirements, design, development, testing, documentation, and deployment.Developed and implemented signal analysis algorithms using Python programming language within the Flask framework.Developed and deployed a machine learning model for real-time patient health prediction by processing radial artery signals.Utilized Numpy, Pandas, and Scikit-learn to perform advanced data analysis on radial artery signals.Wrote Scikit learn-based machine learning algorithms for building POC’s on sample dataset.Implemented machine learning models using PyTorch and Keras to further analyze and interpret intricate patterns within signal data.Worked on generating pre-signed URLs for the transaction documents in AWS S3.Used AWS Simple Notification Service (SNS) to publish any validation failure message to the respective downstream systems.Participated in daily SCRUM meetings in an AGILE environment.Translated application use cases, user stories, and UI application into dynamic web applications.Used Flask to develop RESTful web services. Integrated security with OAuth2 for RESTful API using Python, tested it, and documented the API using Postman.Wrote Unit Tests using pytest and unit test.Used third-party packages like logging for efficient logging of application activities.Worked with MySQL validations to handle error messages at database level.Created and modified Stored Procedure to reduce the network traffic between application and MySQL server.Used AWS S3 Bucket for uploading and storing static images and hosted the web application on AWS creating EC2 Instances.Design and created API documentation using Swagger.

Involved in Full Cycle of Software development from Analysis through Design, Development, Integration, and testing phases.Designed and implemented well-structured Python code using object-oriented principles, ensuring readability, reusability, and testability.Designed and implemented robust and interactive web applications using Python frameworks (Django, Flask).Implemented a robust token lifecycle management system, including token generation, expiration, and revocation strategies, to enhance security and ensure secure user authentication.Designed and implemented token-refresh mechanisms, extending user sessions while maintaining a balance between security and user convenience.Designed and implemented Django models and views to serve as the backbone for RESTful endpoints, facilitating smooth data transmission between the frontend and MySQL database.