Allen Fletcher

- Designed and implemented end-to-end AI pipelines for CBCT reconstruction, integrating imaging physics with deep
learning models.
- Developed 3D deep learning architectures for volumetric reconstruction directly from raw X-ray projections.
- Improved image quality through AI-based denoising and artifact reduction, targeting near-CT fidelity for clinical
usability.
- Incorporated geometric modeling and reconstruction theory into neural architectures, enabling physically
consistent outputs.
- Built AI-assisted anatomical detection systems to support diagnostic workflows and treatment planning.
- Optimized reconstruction workflows to reduce processing time and improve system efficiency (qualitative improvement;
no verified metric available).

- Developed and deployed transformer-based NLP models, improving classification accuracy by 18%.
- Reduced inference latency by 35% using ONNX and TensorRT optimization.
- Designed and deployed retrieval-augmented generation (RAG) pipelines for scalable knowledge systems.
- Containerized and served models using Docker and FastAPI, enabling production-grade deployment.