Sai Ram Lankasairam

Designed and deployed deep learning-based forecasting models using LSTM and Prophet for demand prediction. Built an MLOps pipeline using AWS Sagemaker, Airflow, and MLflow for model lifecycle management. Implemented data versioning and tracking using DVC and GitHub Actions for reproducible experiments. Automated model retraining workflows using Airflow DAGs triggered by sales data updates. Applied feature engineering to include seasonality, competitor price trends, and macroeconomic indicators. Integrated real-time inference APIs into production via FastAPI and Dockerized services. Conducted EDA and anomaly detection using Pandas, Seaborn, and Isolation Forest. Deployed dashboards with Power BI and Looker; set up Kubernetes clusters with Helm; integrated model monitoring with Prometheus and Grafana.

Developed NLP models using BERT and LSTM to analyze and classify customer sentiment from reviews and chat logs. Created data pipelines for ingestion, preprocessing, and tokenization using Airflow and Kubeflow Pipelines. Deployed NLP models to GCP Vertex AI with APIs for real-time predictions. Used MLflow for model versioning and experiment logging. Implemented automated retraining triggered by new data in GCS. Built Tableau dashboards for executive sentiment trend monitoring. Optimized inference time with TensorFlow Serving on GPU-enabled Kubernetes. Applied feature store management for reusable embeddings and pre-processed tokens. Integrated CI/CD with GitLab CI for zero-downtime deployment. Managed preprocessing scripts with Pandas/NumPy and spaCy. Deployed monitoring with Prometheus; collaborated in Agile sprints.

Designed and deployed predictive maintenance models using Random Forest and Gradient Boosting to detect equipment failures early. Built and automated end-to-end ML pipelines using Airflow and MLflow for version control and model retraining. Containerized ML microservices with Docker and Kubernetes in a hybrid cloud setup. Managed data ingestion from SQL and sensor data sources using Pandas and ETL pipelines. Implemented feature store management to reuse features across models. Monitored model drift and performance with Power BI dashboards. Automated deployment using Jenkins CI/CD with MLflow tracking server. Deployed models on AWS (EC2, S3, Lambda) ensuring scalability. Performed hyperparameter tuning with GridSearchCV, improving accuracy by 15%. Collaborated with data engineers and business analysts on feature engineering, and implemented alerting for data drift with Grafana. Optimizations improved system performance by 25%.