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

I'm Swetha Punati, an AI/ML Engineer with 4+ years of experience delivering production-grade healthcare and financial AI solutions. I specialize in computer vision, NLP, and structured EHR data, building end-to-end ML systems using Python, PyTorch, TensorFlow, OpenCV, and Hugging Face. I thrive in cross-functional settings, collaborating with clinicians and domain SMEs to design compliant, auditable AI solutions. I deploy models in production with Docker, Kubernetes, and cloud services (AWS), ensure HIPAA-compliant PHI handling, and enable clinical decision support through multimodal data pipelines, real-time inference, and automated reporting.…I'm Swetha Punati, an AI/ML Engineer with 4+ years of experience delivering production-grade healthcare and financial AI solutions. I specialize in computer vision, NLP, and structured EHR data, building end-to-end ML systems using Python, PyTorch, TensorFlow, OpenCV, and Hugging Face. I thrive in cross-functional settings, collaborating with clinicians and domain SMEs to design compliant, auditable AI solutions. I deploy models in production with Docker, Kubernetes, and cloud services (AWS), ensure HIPAA-compliant PHI handling, and enable clinical decision support through multimodal data pipelines, real-time inference, and automated reporting.

SWETHA PUNATI

AI Engineer, Data Scientist, AI Developer, +2





I'm Swetha Punati, an AI/ML Engineer with 4+ years of experience delivering production-grade healthcare and financial AI solutions. I specialize in computer vision, NLP, and structured EHR data, building end-to-end ML systems using Python, PyTorch, TensorFlow, OpenCV, and Hugging Face. I thrive in cross-functional settings, collaborating with clinicians and domain SMEs to design compliant, auditable AI solutions. I deploy models in production with Docker, Kubernetes, and cloud services (AWS), ensure HIPAA-compliant PHI handling, and enable clinical decision support through multimodal data pipelines, real-time inference, and automated reporting.…I'm Swetha Punati, an AI/ML Engineer with 4+ years of experience delivering production-grade healthcare and financial AI solutions. I specialize in computer vision, NLP, and structured EHR data, building end-to-end ML systems using Python, PyTorch, TensorFlow, OpenCV, and Hugging Face. I thrive in cross-functional settings, collaborating with clinicians and domain SMEs to design compliant, auditable AI solutions. I deploy models in production with Docker, Kubernetes, and cloud services (AWS), ensure HIPAA-compliant PHI handling, and enable clinical decision support through multimodal data pipelines, real-time inference, and automated reporting.

Available to hire

I’m Swetha Punati, an AI/ML Engineer with 4+ years of experience delivering production-grade healthcare and financial AI solutions. I specialize in computer vision, NLP, and structured EHR data, building end-to-end ML systems using Python, PyTorch, TensorFlow, OpenCV, and Hugging Face.

I thrive in cross-functional settings, collaborating with clinicians and domain SMEs to design compliant, auditable AI solutions. I deploy models in production with Docker, Kubernetes, and cloud services (AWS), ensure HIPAA-compliant PHI handling, and enable clinical decision support through multimodal data pipelines, real-time inference, and automated reporting.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Work Experience

Software Engineer – AI/ML at Johnson & Johnson

September 1, 2024 - Present

Captured surgical video streams via platform APIs and stored in encrypted AWS S3 with KMS; enforced HIPAA-compliant PHI handling with IAM RBAC. Integrated Epic EHR metadata via FHIR REST APIs into PostgreSQL for model context and surrogate labeling. Built CVAT labeling pipelines for video segmentation; exported COCO-style JSON and synchronized timestamps with EHR fields. Performed frame extraction, augmentation, and step segmentation using NumPy/OpenCV, reducing annotation overhead. Trained action recognition and instrument detection models with PyTorch; applied Mixed Precision Training on NVIDIA V100 GPUs to reduce training time. Deployed multi-model inference via NVIDIA Triton; implemented postoperative summary generation with Llama 2 hosted on Hugging Face and created Streamlit dashboards for SME validation. Collected inference metrics with Prometheus and Grafana; documented HIPAA-alignment guidelines.

AI/ML Engineer at Capgemini

January 1, 2020 - August 31, 2022

Developed fraud detection using PyTorch DNNs; reduced false positives by 31% and improved anomaly detection across high-volume transactional environments. Calibrated classification models for class imbalance via SMOTE; improved recall across multiple institutions. Built NLP pipelines with Hugging Face Transformers (BERT) for KYC document analysis, extracting risk entities and sentiment, improving monitoring efficiency. Containerized ML inference services with Docker and deployed on AWS SageMaker Endpoints, enabling A/B testing and canary rollouts. Implemented real-time fraud/NPS scoring with FastAPI, Spark windowed aggregations, and Kafka; latency under 200 ms on AWS Lambda. Used Optuna for hyperparameter tuning; built explainable dashboards with SHAP; standardized ML lifecycle with MLflow and GitHub Actions; drift detection and retraining with reduced manual intervention.