Building production-grade AI & LLM systems that scale.

Comerica, Dallas, Texas, USA

• Engineered fraud detection, risk scoring, and financial forecasting models across 20M+ banking transactions, improving AUC-ROC by 15 points.
• Architected Spark-based data platforms processing 10TB+ of financial data, reducing processing time by 40%.
• Fine-tuned domain-specific LLMs using LoRA/PEFT and Hugging Face, improving classification accuracy by 22%.
• Built enterprise RAG platforms with LangChain, LangGraph, FAISS and Pinecone across 500K+ financial documents, cutting retrieval time by 60%.
• Deployed multi-agent AI systems on AWS SageMaker/EC2/Lambda, eliminating 150+ manual hours weekly.

Dell, Round Rock, Texas, USA

• Built forecasting and optimization models improving forecasting accuracy by 20% across global supply-chain operations.
• Engineered data pipelines processing 15TB+ using Spark, Databricks, Kafka, and Hadoop.
• Used GCP Vertex AI and BigQuery alongside AWS SageMaker, cutting model experimentation cycle time by 30%.
• Architected GenAI/RAG solutions enabling citation-grounded search across 250K+ documents.
• Designed LangGraph-based multi-agent workflows, reducing manual resolution effort by 40%.

South Indian Bank, Hyderabad, India

• Analyzed 50M+ customer and transaction records for risk, lending, and portfolio analytics.
• Developed ML models for fraud detection and loan default prediction, improving prediction accuracy by 18%.
• Built NLP-driven document intelligence platforms with BERT and spaCy, reducing manual review effort by 45%.

GeBBS Healthcare Solutions, Hyderabad, India

• Built predictive analytics across 20M+ EHR, claims, and patient records, improving risk prediction accuracy by 15%.
• Developed NLP clinical intelligence applications to extract insights from physician notes and discharge summaries.
• Delivered HIPAA-compliant analytics and executive dashboards across 15+ healthcare facilities.