Baisa Supriya

● AI-ML Engineer predictive maintenance and industrial intelligence platform using GPT-4, Llama 3, PyTorch, AWS Bedrock, and advanced RAG pipelines, incorporating embeddings and contextual retrieval to reduce unplanned downtime by 25% and improve global asset reliability. ● Architected Python-based, cloud-native GenAI backend services leveraging AWS Bedrock, Amazon Kinesis, vector databases, EKS, LangChain, FastAPI, and REST APIs, enabling scalable LLM-powered microservices and real-time inference for millions of industrial IoT events. ● Executed Agile Scrum delivery for Generative AI systems, including sprint planning, backlog grooming, CI/CD alignment, and cross-functional collaboration, ensuring iterative delivery of LLM APIs, compliance adherence, predictable releases, and stakeholder transparency. ● Ingested structured and unstructured telemetry using Amazon Kinesis, REST APIs, SQL, NoSQL, and Amazon S3, supporting real-time streaming and batch pipelines for GenAI-driven predictive analytics, embeddings at scale, and RAG-based contextual retrieval. ● Processed large-scale industrial datasets using PySpark, Pandas, AWS Glue, and EMR, performing feature engineering, normalization, enrichment, text chunking, and embedding generation to support downstream RAG and Graph-aware retrieval pipelines. ● Stored curated datasets in Amazon Redshift, Amazon S3, and PostgreSQL with indexing, partitioning, and lifecycle policies, enabling high-throughput analytics, historical backtesting, retraining, and enterprise GenAI experimentation. ● Implemented specialized low-latency storage and vector retrieval layers using Amazon ElastiCache (Redis) and vector databases, enabling embedding retrieval, semantic search, knowledge-linked context, and near–real-time LLM inference for industrial workloads. ● Selected and evaluated foundation models and LLMs including GPT-4, Llama 3, Claude, Hugging Face Transformers, and Amazon Titan via AWS Bedrock.

Delivered a scalable enterprise AI platform on Azure ML, AKS, Azure Functions, and API Management, automating NLP, computer vision, and predictive analytics to improve decision accuracy by 22%. Designed distributed ML architectures integrating Event Hubs, Azure Data Factory, Synapse Analytics, and Azure ML pipelines to support hybrid real-time and batch inference workloads. Built robust data ingestion and ETL pipelines using Event Hubs, Kafka connectors, Azure Databricks (Spark), and Data Lake Gen2, enabling reliable feature engineering and model-ready datasets. Developed and optimized ML and deep learning models (XGBoost, LightGBM, CNNs, LSTMs, transformer-based NLP) for fraud detection, credit scoring, image classification, and recommendation systems. Implemented end-to-end MLOps with Docker, Azure Container Registry, AKS, Azure ML Pipelines, and Feature Store, ensuring reproducible training, consistent online/offline features, and governed deployments. Led Agile Scrum delivery, collaborating closely with product managers and data scientists, applying rigorous experimentation (A/B testing, cross-validation) to deliver measurable, production-ready AI outcomes.

Delivered scalable healthcare ML solutions using Python, TensorFlow, PyTorch, Spark, improving statewide predictive health models and enabling data-driven clinical and policy decisions. Designed distributed ML pipelines with Spark ETL, Hadoop/S3, AWS EMR/EKS, TensorFlow training, and batch/REST inference supporting real-time clinical analytics. Built and engineered large-scale healthcare datasets (EHR, claims, registries) using PySpark and Python, enabling longitudinal patient modeling and regulatory reporting. Applied advanced ML & NLP techniques (logistic regression, gradient boosting, clustering, CNNs, LSTMs, transformers, TF-IDF) for prediction, segmentation, and personalized care analytics. Optimized models via hyperparameter tuning, cross-validation, GPU acceleration, and explainability checks to ensure robust, low-latency clinical inference. Containerized and deployed ML workflows using Docker, Kubernetes (EKS), CI/CD, operating in Agile Scrum environments with cross-functional healthcare teams.