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

Hi, I’m Sanjana Dadi—a seasoned AI engineer with over a decade of experience designing, building, and deploying scalable ML and Generative AI systems across financial services, healthcare, and insurance. I specialize in end-to-end ML pipelines, LLM-based architectures, retrieval-augmented generation, and policy-driven governance for enterprise AI. I thrive on turning complex data problems into reliable, production-ready solutions. My strengths include prompt engineering, parameter-efficient fine-tuning (LoRA/PEFT), observability, and performance optimization, all backed by hands-on experience with modern MLOps and cloud platforms.…Hi, I’m Sanjana Dadi—a seasoned AI engineer with over a decade of experience designing, building, and deploying scalable ML and Generative AI systems across financial services, healthcare, and insurance. I specialize in end-to-end ML pipelines, LLM-based architectures, retrieval-augmented generation, and policy-driven governance for enterprise AI. I thrive on turning complex data problems into reliable, production-ready solutions. My strengths include prompt engineering, parameter-efficient fine-tuning (LoRA/PEFT), observability, and performance optimization, all backed by hands-on experience with modern MLOps and cloud platforms.

Sanjana Dadi





Hi, I’m Sanjana Dadi—a seasoned AI engineer with over a decade of experience designing, building, and deploying scalable ML and Generative AI systems across financial services, healthcare, and insurance. I specialize in end-to-end ML pipelines, LLM-based architectures, retrieval-augmented generation, and policy-driven governance for enterprise AI. I thrive on turning complex data problems into reliable, production-ready solutions. My strengths include prompt engineering, parameter-efficient fine-tuning (LoRA/PEFT), observability, and performance optimization, all backed by hands-on experience with modern MLOps and cloud platforms.…Hi, I’m Sanjana Dadi—a seasoned AI engineer with over a decade of experience designing, building, and deploying scalable ML and Generative AI systems across financial services, healthcare, and insurance. I specialize in end-to-end ML pipelines, LLM-based architectures, retrieval-augmented generation, and policy-driven governance for enterprise AI. I thrive on turning complex data problems into reliable, production-ready solutions. My strengths include prompt engineering, parameter-efficient fine-tuning (LoRA/PEFT), observability, and performance optimization, all backed by hands-on experience with modern MLOps and cloud platforms.

Available to hire

Hi, I’m Sanjana Dadi—a seasoned AI engineer with over a decade of experience designing, building, and deploying scalable ML and Generative AI systems across financial services, healthcare, and insurance. I specialize in end-to-end ML pipelines, LLM-based architectures, retrieval-augmented generation, and policy-driven governance for enterprise AI.

I thrive on turning complex data problems into reliable, production-ready solutions. My strengths include prompt engineering, parameter-efficient fine-tuning (LoRA/PEFT), observability, and performance optimization, all backed by hands-on experience with modern MLOps and cloud platforms.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Work Experience

AI Engineer at Radian Group

June 1, 2022 - December 1, 2025

Led end-to-end design and deployment of large language model–based solutions using Hugging Face Transformers, improving inference reliability by 30% through structured evaluation frameworks, benchmark comparisons, optimized tokenization strategies, and reproducible experiment pipelines aligned with enterprise AI governance standards. Architected OpenAI API integrations across customer-facing AI applications, reducing response latency by 25% and token usage costs by 22% through prompt compression, caching, intelligent retry logic, and structured monitoring across high-volume production environments. Designed scalable retrieval-augmented generation pipelines with Pinecone vector databases, improving retrieval precision by 20% and reducing hallucination rates by 18% through optimized embedding strategies and chunking methodologies. Implemented robust LangChain orchestration frameworks incorporating tool-calling, memory layers, and retrieval workflows, improving system reliability by 35%

Senior Machine Learning Engineer at Highmark Health

January 1, 2021 - May 1, 2022

Led development and optimization of large-scale ML workloads on Advanced Databricks, reducing model training time by 42% through cluster right-sizing, adaptive autoscaling, memory tuning, and optimized feature engineering pipelines that improved reliability and cost efficiency across multiple teams. Architected scalable ML data pipelines and feature stores in Snowflake, ensuring full consistency between training and inference environments while improving data freshness by 25% through automated ingestion workflows and governance-driven schema controls. Designed, trained, and deployed production ML models using Azure ML pipelines and model registry, reducing deployment errors by 35% and accelerating release cycles by 40% through automated validation, versioning, and rollback strategies. Deployed containerized ML workloads on Kubernetes clusters, achieving 99.8% availability by implementing resource limits, autoscaling policies, secure RBAC configurations, centralized logging, and proacti

Machine Learning Engineer at The Hanover Insurance Group

January 1, 2019 - December 1, 2020

Designed and implemented end-to-end machine learning models using TensorFlow for structured and semi-structured insurance datasets, improving prediction accuracy by 13% through advanced feature engineering, balanced sampling strategies, and structured cross-validation aligned with enterprise risk modeling objectives. Developed distributed training pipelines using PyTorch with optimized data loaders and GPU scheduling strategies, reducing overall model training time by 35% while maintaining convergence stability across multi-node cloud environments. Built and optimized gradient boosting models using XGBoost and LightGBM, achieving a 15% lift in predictive precision over baseline logistic regression models through hyperparameter tuning and ensemble techniques. Containerized machine learning workflows using Docker, creating reproducible runtime environments that reduced deployment inconsistencies by 40% across development, staging, and production environments. Designed and orchestrated ML

Hadoop Developer at Trianz

November 1, 2016 - December 1, 2018

Engineered large-scale big data pipelines using Apache Spark, Hive, and Sqoop to process approximately 5 terabytes of structured telecom and operational datasets daily, improving ETL throughput by 25% through partition optimization, executor tuning, and workflow orchestration improvements. Designed and implemented distributed Spark transformations and Hive queries to clean, aggregate, and enrich high-volume datasets, reducing data preparation latency by 22% and improving downstream reporting consistency for analytics teams. Developed reusable Python-based data processing scripts using Pandas and NumPy, reducing manual reporting effort by 30% through automation of preprocessing logic and structured validation routines. Built baseline machine learning models using Scikit-learn for segmentation and churn analysis use cases, improving classification performance by 18% through structured feature engineering and cross-validation strategies. Implemented Logistic Regression models for binary c

Data Consultant at ValueLabs

June 1, 2014 - March 1, 2016

Developed complex SQL queries using MySQL and SQL Server to extract and aggregate operational datasets, reducing manual data preparation effort by 25% through reusable query templates and indexing optimization. Designed structured Excel-based reporting frameworks using pivot tables and Power Query, reducing report preparation time by 30% while improving reporting consistency. Built interactive Tableau dashboards improving dashboard refresh performance by 25% through optimized extracts and structured visual design. Automated repetitive data cleaning and transformation tasks using Python and Pandas, reducing analyst workload by 20% and improving repeatability. Performed descriptive statistical analysis improving reporting clarity and stakeholder decision confidence. Implemented structured data validation rules improving dataset accuracy by 15%. Optimized SQL query performance improving execution time by 18%. Standardized recurring reporting templates reducing inconsistencies by 20%. Coll