T
Teja Pravallika Nandamuri
About
Detail
Irving, Texas, United States
Résumé
Jobs
verified_user
0% verified
Research assistant(voluntary)Northern Arizona UniversityDec 2024 - Jun 2025 (7 months)Built an end-to-end adversarial evaluation framework for multimodal LLMs, including data ingestion, attack generation, metric logging, and automated failure analysis. • Fine-tuned and evaluated a 4B-parameter vision-language model, reducing attack success rates by 18% through adversarial retraining and systematic error analysis. • Integrated robustness benchmarks into reusable Python modules to support iterative retraining and regression testing across model versions. • Structured the evaluation framework as reusable, configurable Python modules to support automated testing and integration into future model development workflows- AI/ML EngineerathenahealthJun 2024 - Current (2 years 3 months)• Architected an AI-Powered Ambient Clinical Documentation platform, automating physician-patient conversation capture and clinical note generation across 120K+ encounters, improving documentation turnaround efficiency by 27%. • Engineered scalable audio-processing pipelines using Apache Spark, Apache Airflow, and Google Cloud, processing 540GB+ weekly clinical transcription and encounter datasets for AI-driven documentation workflows. • Developed contextual NLP pipelines using PyTorch and Hugging Face Transformers, analyzing 2.4M+ clinical utterances to improve entity extraction, summarization accuracy, and SOAP-note generation consistency. • Implemented FAISS vector indexing for semantic retrieval across 950K+ clinical conversations, acce
- AI/ML EngineerAccentureMay 2021 - Jul 2023 (2 years 3 months)• Architected a Generative AI-powered Intelligent Document Processing platform using Python, SQL, and AWS SageMaker, automating extraction and validation across 5.4M+ financial and compliance documents annually. • Developed predictive AI models using XGBoost and Scikit-learn for document prioritization, fraud flagging, and exception handling, improving processing accuracy by 19% across enterprise compliance workflows. • Engineered NLP pipelines with Hugging Face Transformers and BERT, automating OCR correction, entity extraction, semantic classification, and intelligent summarization across 620K+ customer documents annually. • Implemented statistical validation using A/B testing, chi-square tests, uplift modeling, and confidence intervals,
Education
verified_user
0% verified
- Bachelor’s In Computer ScienceIndiaJul 2018 - Jun 2022 (4 years)
- UMaster’s In Computer ScienceUSAAug 2023 - Dec 2024 (1 year 5 months)