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

Hi, I’m Harshitha Saripalli, a machine learning engineer with more than five years of hands-on experience building production ML pipelines across financial services, healthcare claims, and user-behavior systems. I’m passionate about turning data into reliable models and scalable services that end users rely on every day. I’ve led end-to-end work from feature design and model training to real-time inference, monitoring, and deployment. My strengths lie in practical ML engineering—improving accuracy, latency, and data quality—using Python, PyTorch, scikit-learn, Docker, Kubernetes, and modern inference stacks to deliver business impact without overreliance on theory.…Hi, I’m Harshitha Saripalli, a machine learning engineer with more than five years of hands-on experience building production ML pipelines across financial services, healthcare claims, and user-behavior systems. I’m passionate about turning data into reliable models and scalable services that end users rely on every day. I’ve led end-to-end work from feature design and model training to real-time inference, monitoring, and deployment. My strengths lie in practical ML engineering—improving accuracy, latency, and data quality—using Python, PyTorch, scikit-learn, Docker, Kubernetes, and modern inference stacks to deliver business impact without overreliance on theory.

Harshitha Saripalli

Data Scientist, AI Engineer, AI Developer, +2





Hi, I’m Harshitha Saripalli, a machine learning engineer with more than five years of hands-on experience building production ML pipelines across financial services, healthcare claims, and user-behavior systems. I’m passionate about turning data into reliable models and scalable services that end users rely on every day. I’ve led end-to-end work from feature design and model training to real-time inference, monitoring, and deployment. My strengths lie in practical ML engineering—improving accuracy, latency, and data quality—using Python, PyTorch, scikit-learn, Docker, Kubernetes, and modern inference stacks to deliver business impact without overreliance on theory.…Hi, I’m Harshitha Saripalli, a machine learning engineer with more than five years of hands-on experience building production ML pipelines across financial services, healthcare claims, and user-behavior systems. I’m passionate about turning data into reliable models and scalable services that end users rely on every day. I’ve led end-to-end work from feature design and model training to real-time inference, monitoring, and deployment. My strengths lie in practical ML engineering—improving accuracy, latency, and data quality—using Python, PyTorch, scikit-learn, Docker, Kubernetes, and modern inference stacks to deliver business impact without overreliance on theory.

Available to hire

Hi, I’m Harshitha Saripalli, a machine learning engineer with more than five years of hands-on experience building production ML pipelines across financial services, healthcare claims, and user-behavior systems. I’m passionate about turning data into reliable models and scalable services that end users rely on every day.

I’ve led end-to-end work from feature design and model training to real-time inference, monitoring, and deployment. My strengths lie in practical ML engineering—improving accuracy, latency, and data quality—using Python, PyTorch, scikit-learn, Docker, Kubernetes, and modern inference stacks to deliver business impact without overreliance on theory.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Language

English

Fluent

Work Experience

AI/ML Engineer at Intuit

August 1, 2024 - Present

Built an NLP pipeline using spaCy and transformer embeddings to clean invoice text and normalize vendor descriptions, boosting category-mapping accuracy from 78% to 90% across 1.5M+ transactions. Trained a transaction anomaly model with PyTorch and XGBoost to flag duplicate charges and reconciliation mismatches, reducing bookkeeping errors by over $3.4M annually. Developed a RAG workflow with custom financial embeddings and vector search to surface accounting rules and policy snippets inside internal tools, cutting agent lookup time by 35%. Designed a cash-flow forecasting pipeline using Prophet and scikit-learn with engineered bank-feed features, improving short-term forecast precision by 19% for small businesses. Served models via FastAPI and Redis with async inference and caching, reducing p95 latency from 530 ms to 250 ms.

AIML Engineer Intern at Humana

January 1, 2024 - July 1, 2024

Analyzed EMR and claims datasets and built a Gradient Boosting model to score readmission risk, improving high-risk patient capture from 70% to 84%, helping care teams prioritize outreach. Designed an XGBoost anomaly scorer to spot unusual utilization and repeated procedures in Medicare Advantage claims, lowering audit review noise by 22%. Processed physician notes with spaCy to extract diagnosis terms and normalize text for feature creation; raised ICD keyword-match accuracy from 74% to 88%. Automated data prep on AWS Glue and Redshift, transforming 100M+ claim rows for downstream modeling; shortened preprocessing cycles from 9 hours to 3 hours.

Machine Learning Engineer at VIVMA Software Inc.

May 1, 2018 - September 1, 2022

Engineered batch training pipelines in Python and scikit-learn to retrain customer-behavior models on weekly event logs, lifting prediction stability by 17%. Implemented a TensorFlow inference service to score user sessions in real time, exposing predictions through REST APIs; cut response latency from 420 ms to 180 ms with model quantization. Designed feature-generation jobs in SQL and Pandas to blend clickstream, transaction, and session metadata; improved model AUC from 0.71 to 0.82 after adding temporal features. Set up monitoring dashboards using Prometheus and Python checks to track drift, stale features, and traffic anomalies; reduced unexpected model failures by 28%. Implemented CI/CD with GitHub Actions and containerized builds; developed data-validation tests with PyTest, catching schema shifts and missing fields;