Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

AI & ML Engineer with 4 years of hands-on experience building and shipping production AI systems across financial services, enterprise SaaS, and consulting. Focused on real-world performance across LLM applications, Retrieval-Augmented Generation (RAG), NLP pipelines, and end-to-end ML infrastructure. Driven measurable outcomes including 30%+ gains in model accuracy, 35% reductions in manual processing overhead, and sub-150ms inference latency in high-throughput systems. Experienced across the full ML lifecycle—from data ingestion to monitored production deployment on AWS and Azure.…AI & ML Engineer with 4 years of hands-on experience building and shipping production AI systems across financial services, enterprise SaaS, and consulting. Focused on real-world performance across LLM applications, Retrieval-Augmented Generation (RAG), NLP pipelines, and end-to-end ML infrastructure. Driven measurable outcomes including 30%+ gains in model accuracy, 35% reductions in manual processing overhead, and sub-150ms inference latency in high-throughput systems. Experienced across the full ML lifecycle—from data ingestion to monitored production deployment on AWS and Azure.

Durga Daruvuri

AI Engineer, Data Scientist, Data Engineer, +2





AI & ML Engineer with 4 years of hands-on experience building and shipping production AI systems across financial services, enterprise SaaS, and consulting. Focused on real-world performance across LLM applications, Retrieval-Augmented Generation (RAG), NLP pipelines, and end-to-end ML infrastructure. Driven measurable outcomes including 30%+ gains in model accuracy, 35% reductions in manual processing overhead, and sub-150ms inference latency in high-throughput systems. Experienced across the full ML lifecycle—from data ingestion to monitored production deployment on AWS and Azure.…AI & ML Engineer with 4 years of hands-on experience building and shipping production AI systems across financial services, enterprise SaaS, and consulting. Focused on real-world performance across LLM applications, Retrieval-Augmented Generation (RAG), NLP pipelines, and end-to-end ML infrastructure. Driven measurable outcomes including 30%+ gains in model accuracy, 35% reductions in manual processing overhead, and sub-150ms inference latency in high-throughput systems. Experienced across the full ML lifecycle—from data ingestion to monitored production deployment on AWS and Azure.

Available to hire

AI & ML Engineer with 4 years of hands-on experience building and shipping production AI systems across financial services, enterprise SaaS, and consulting. Focused on real-world performance across LLM applications, Retrieval-Augmented Generation (RAG), NLP pipelines, and end-to-end ML infrastructure.

Driven measurable outcomes including 30%+ gains in model accuracy, 35% reductions in manual processing overhead, and sub-150ms inference latency in high-throughput systems. Experienced across the full ML lifecycle—from data ingestion to monitored production deployment on AWS and Azure.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Beginner

Beginner

Language

English

Advanced

Work Experience

AI & ML Engineer at Morgan Stanley

February 1, 2026 - Present

Owned end-to-end RAG system architecture using hybrid dense + sparse retrieval, cutting response latency and pushing accuracy up ~30% through custom chunking and reranking. Reduced manual processing effort by 35% by leading design and deployment of NLP classification systems handling 100K+ records monthly. Cut model release cycle time by 25% by architecting standardized ML pipelines across teams (from data ingestion through production inference). Built and owned agentic AI workflows for multi-step document reasoning, improving complex query accuracy and consistency at scale. Reduced inference cost and latency via caching strategies, prompt tuning, and retrieval optimization; maintained production-stable and low-overhead performance.

AI & ML Engineer at Salesforce

August 1, 2024 - January 31, 2026

Improved classification and anomaly detection accuracy by 18% by owning XGBoost and Random Forest model development across enterprise systems. Cut data preprocessing time by 30% by building automated ETL and feature engineering pipelines in Python and Airflow. Delivered sub-150ms inference latency in production by deploying real-time ML APIs using FastAPI, Docker, and Kubernetes. Built structured evaluation frameworks benchmarking accuracy, latency, and business constraints to improve model selection. Designed full-stack ML workflows covering ingestion, training, and deployment for scalable enterprise AI delivery.

AI & ML Engineer at Tata Consultancy Services

June 1, 2022 - May 31, 2023

Built fraud detection and churn prediction models improving risk identification accuracy by 20% across enterprise datasets. Engineered distributed PySpark pipelines across multi-million record datasets, improving processing efficiency and reliability at scale. Developed LSTM-based time series forecasting models, outperforming traditional statistical baselines by 15% on key benchmarks. Delivered real-time ML KPI visibility to business stakeholders through Power BI and Tableau dashboards. Implemented SQL-based validation frameworks enforcing data consistency and accuracy across production ML pipelines.

AI & ML Engineer at Genpact

June 1, 2021 - May 31, 2022

Built spaCy-based NLP pipelines for grievance classification, improving ticket prioritization efficiency by 30% across teams. Developed classification and clustering models supporting demand forecasting and resource allocation across enterprise systems. Ran A/B tests and controlled experiments to validate model performance prior to production deployment decisions. Designed scalable data pipelines improving system reliability and data availability for ML workflows. Documented ML systems in Confluence to improve reproducibility and cross-team handoffs.