Brian User

Designed and deployed end-to-end LLM and retrieval pipelines (AdGPT) using Haystack, PGVectorStore, and FastAPI/gRPC, improving natural-language access to building data and automating GraphQL query generation. Built chatbot integrations, observability, and feedback loops to increase LLM accuracy and production reliability.

Built a healthcare Q&A system combining React frontend and FastAPI backend, integrating fine-tuned BERT/GPT/LLaMA models with RAG and deploying via Docker to deliver real-time responses and improved latency. Implemented PostgreSQL and Redis for storage and caching.

Designed and deployed AI-driven chatbots and NLP pipelines (intent detection, entity recognition, sentiment) using spaCy, NLTK, and BERT; built Django REST APIs and integrated AWS services (EC2, S3) to support production chat systems.

Developed medical ML models for disease prediction and imaging using scikit-learn, TensorFlow/Keras and CNN architectures (GoogLeNet, VGG16), and built NLP pipelines to extract clinical entities from notes, improving diagnostic model performance.

Developed image classification and object detection prototypes using MATLAB and CNN frameworks, preprocessing and annotating medical imaging datasets to improve biomarker detection accuracy and reduce false negatives.

Completed a Master's degree in Computer Science and Engineering with coursework and projects focused on advanced algorithms, machine learning, and software engineering.

Completed a Bachelor's degree in Computer Science and Engineering with foundational training in programming, data structures, and systems development.