Mamidi Sri Teja

As an Applied AI Engineer, I am driven by a deep curiosity and ambition to master complex challenges, constantly seeking innovative solutions to real-world problems. My diligent approach ensures high-quality outcomes, and I thrive in collaborative environments where collective effort leads to groundbreaking advancements. I am dedicated to continuously refining my expertise and pushing the boundaries of AI to deliver impactful results.

Experience

Applied AI Engineer - Claims Intelligence Platform

Deloitte · Apr 2025 – Present

Not yet confirmed

• Built and maintained production RAG pipelines (ChromaDB vector retrieval, FlashRank cross-encoder reranking) supporting LLM-assisted compliance reporting and claim decision support monitored pipeline performance via CloudWatch and Tableau dashboards.
• Built anomaly detection spanning both infrastructure telemetry (AWS CloudWatch statistical baselines across 5 claims microservices, 10k+ daily transactions) and the fraud detection/claims estimation models themselves (monitoring for prediction drift and anomalous outputs, carried over from Carelon, with MLflow experiment tracking) combined impact: 40% reduction in MTTD and 30% reduction in P1 escalations.
• Supported the document processing pipeline automated text extraction, document classification, and key information extraction from scanned claim documents in the intake workflow.
• Delivered 15+ RCA-driven fixes across priority claims services, reducing recurring defects by 30% and sustaining zero critical post-release defects across Q2-Q3 2025.
• Built and maintained MLOps and CI/CD pipelines for claims AI modules with automated regression testing, sustaining zero production downtime across releases.

C

Software Engineer - ML-Augmented Claims Platform

Carelon Global Solutions · Nov 2023 – Mar 2025

Not yet confirmed

• Wrote feature engineering and model training code for the fraud detection model (XGBoost classifier) deployed with the team to AWS SageMaker for real-time inference, with model artifacts stored in S3.
• Wrote feature engineering and model training code for the claims cost and time estimation models (Gradient Boosting Regressor, XGBoost Regressor) integrated into the adjudication workflow and deployed via SageMaker.
• Contributed to integrating the document processing pipeline (Tesseract OCR, spaCy, BERT NER) into the claims intake system alongside the fraud and estimation models.
• Monitored production model metrics and business KPIs across all three systems; maintained the automated retraining pipeline that refreshes models against claims data drift.

What drives their work

The Builder

Turns concepts into robust, deployed systems.

Mamidi's superpower

You excel at bringing complex AI/ML projects to life, ensuring they are built, deployed, and monitored effectively.

How Mamidi works

You would thrive in teams focused on developing and deploying scalable AI/ML solutions, especially those that encourage end-to-end ownership and continuous learning.