🚀 Experienced Senior Machine Learning Engineer | Transforming Data into Compelling ProductsAs a seasoned ml engineer with a track record of success, I specialize in developing cutting-edge solutions that harness the power of advanced analytics, machine learning, and AI. Currently on a dynamic team at Aware, I have spearheaded the creation of a retrieval augmented generation system, utilizing a self-hosted large language model and vector database in Kubernetes. This innovative system empowers efficient, accurate, and meaningful exploration of employee collaboration data.In my role, I delivered a responsible and efficacious large language model-powered summarization engine, elevating customer retention and driving new business with an industry-leading content summarization solution. Leveraging PyTorch, I developed a mixture density network that predicts daily health and sentiment distributions, offering state-of-the-art insights into the well-being of globally recognized brands based on over 1 billion observations.My expertise extends to implementing a multilingual trending themes pipeline using distributed computing, processing over 50 million content objects. This initiative not only increased data volume by a factor of 5 but also decreased training time by a factor of 20. Additionally, I mentored a data scientist intern, guiding a project to model customer lifecycle and understand the sales funnel, resulting in substantial increases in net revenue.Previously at Express, I enhanced customer lifetime value prediction accuracy through a custom neural network architecture in TensorFlow, driving more effective marketing segmentation. I also developed a markdown optimization model that significantly increased net margin by over $2 million annually.In my tenure at International Information Associates, contracted to the US DOE (OSTI), I played a crucial role as a Data Scientist. My accomplishments include implementing an RNN and Word2Vec pipeline to categorize millions of unstructured documents, improving the efficiency and accuracy of nuclear information classification. I collaborated with the Defense Technical Information Center on the algorithmic detection of sensitive data and created an automated curation system for a semantic thesaurus using Word2Vec embeddings.https://github.com/louisdavidsmith