Looking for a job

Sarmad Khan

@sarmadkhan4

Senior Data Scientist at Elevance Health building AI agents that cut policy review time by 56%.

At Elevance Health, I develop LLM-powered products for healthcare market and policy intelligence. My agents automate extraction, summarization, and Q&A across policy and competitor documents, reducing manual review time by 56%.

I’ve improved retrieval accuracy by 30% with RAG pipelines and built evaluation guardrails that reduced GenAI policy violations by 35%. Earlier at Capgemini, I developed fraud detection and credit risk models that reduced false positives by 28%, and at Fiserv I helped increase transaction data processing throughput by 30%.

Experience

Work history, roles, and key accomplishments

Current

Senior Data Scientist, AI & Market Intelligence

Current

Elevance Health

Feb 2020 - Present (6 years 8 months)

Lead development of LLM-powered data products and agents for healthcare market and policy intelligence, reducing manual review time by 56% and improving retrieval accuracy by 30%. Build scalable data pipelines and evaluation frameworks to ensure accuracy and compliance.

Python SQL PySpark Airflow GPT 3.5 Flan T5 Vertex AI Lang Chain Faiss Sentence Transformers RAG MLFlow Weights Biases Tesseract OCR OpenCV LayoutLMv3 Streamlit QuickSight AWS Docker fastAPI HIPAA WCAG 2.2

Data Scientist, AI/ML

Capgemini

Oct 2017 - Dec 2019 (2 years 2 months)

Built fraud detection and credit risk models using XGBoost and deep learning, reducing false positives by 28%. Fine-tuned LLMs for financial document classification and deployed models via MLOps pipelines on GCP and Databricks.

Python SQL PySpark XGBoost scikit learn TensorFlow Hugging Face Transformers BERT GPT SHAP LIME GCP Databricks Django Matplotlib SeaBorn Plotly Basel III GDPR

Software Developer

Fiserv

Sep 2015 - Aug 2017 (1 year 11 months)

Built web and mobile applications with Python, Java, and JavaScript, and developed scalable backend services and data pipelines using AWS, Kafka, and MongoDB, increasing transaction processing throughput by 30%.