DheeraJ Kandagatla

Built an end-to-end ML and GenAI platform for loan risk: engineered loan default risk prediction with AWS SageMaker (22% accuracy lift) and deployed zero-latency inference via AWS Lambda. Architected RAG systems for natural-language querying over internal financial datasets using LangChain with FAISS and Snowflake for context-aware retrieval. Developed LLM-powered automation using Claude API and OpenAI for report generation, compliance summarization, and anomaly flagging. Implemented embedding-based semantic search for financial policies/regulations with Pinecone to reduce analyst research time (~40%). Created CI/CD-driven model retraining workflows with GitHub Actions and Jenkins, and supported regulatory governance using model cards, bias assessments, and collaboration with compliance teams. Visualized KPIs, risk distributions, and drift metrics with Power BI.

Developed healthcare equipment failure prediction models that reduced unplanned downtime by 30% and improved preventive maintenance scheduling accuracy across hospital fleets. Built time-series anomaly detection using ARIMA and LSTM on IoT sensor data to identify faults up to 48 hours before failures. Designed scalable preprocessing pipelines with Python and Apache Spark for high-frequency sensor streams from 1,000+ devices. Added NLP/interpretability workflows (including SHAP) and delivered real-time failure alerts through REST APIs using Flask and AWS Lambda. Accelerated training with TensorFlow GPU to cut training time by 60% and automated HIPAA-compliant ETL using Airflow DAGs.

Created automated ETL workflows in Python and Pandas to process clinical datasets, reducing data preparation time by 35% and accelerating downstream analytics. Built disease progression prediction models with Scikit-learn, improving recall by 18% using targeted feature engineering and hyperparameter tuning. Implemented NLP pipelines (TF-IDF and sentiment classification) on patient feedback to surface actionable insights for care quality improvements. Developed Power BI dashboards for patient trends, treatment effectiveness, and cost metrics for clinical and administrative stakeholders.