Charles Rice — Senior Data Scientist

Charles has a rich and varied background that blends mathematics, machine learning, and deep programming experience to tackle challenging problems in creative ways.

After earning his Bachelor's in Mathematics from Wayne State University, Charles pursued advanced studies in machine learning at Portland State University. There, he specialized in neural networks, reinforcement learning, and genetic algorithms, contributing to research projects involving evolvable hardware and automated stock trading.

With over 30 years of programming experience, Charles has demonstrated proficiency in a vast array of tools and programming languages. Charles is well-versed in Python, Pandas, NumPy, spaCy, TensorFlow, Keras, PyTorch, Scikit-Learn, SQL, and various other data science and machine learning libraries. This extensive skill set has been honed through his involvement in projects for renowned companies such as Ford Motor Company, Btrieve, Tektronix, Nike, Xerox, and Hewlett-Packard. Charles has frequently assumed leadership roles, contributing technical and strategic insight to project teams.

Data-Centric AI

I am a firm believer in the data-centric AI paradigm, which emphasizes the strategic management and optimization of data to drive intelligent decision-making and enhance model performance. In my work, I prioritize placing data at the heart of AI development, recognizing that well-curated, high-quality data is far more valuable than sheer volume. By focusing on relevance and alignment with specific project goals, I ensure that the AI models I build are not only accurate but also adaptive, evolving with each new data point to continuously improve performance. This approach underpins my commitment to creating robust, reliable solutions that deliver tangible business impact.

Data Science

With a passion for extracting meaningful insights from data, I’ve gained hands-on experience in diverse areas like neural networks, NLP, machine learning, and data analysis. My projects have ranged from modeling tabular data to crafting effective visualizations, always with a focus on leveraging cutting-edge libraries and frameworks to deliver real-world, actionable solutions.

Natural Language Processing (NLP)

I have developed NLP solutions that handle tasks such as topic modeling, information extraction, and integrating Retrieval-Augmented Generation (RAG) with open-source LLMs to improve model outputs.

My experience includes working with tools like spaCy, Sentence Transformers, OpenAI API and Ollama, using them to develop models that handle real-world data and solve specific challenges. I worked on a project where I used NLP for topic modeling to study medical cannabis effects, analyzing large volumes of unstructured text data to uncover trends. Additionally, I have leveraged NLP techniques for extracting structured data from product descriptions. I also explored the potential of large language models (LLMs) for automating customer segmentation, enhancing marketing efforts by improving the accuracy of customer categorization. My experience also includes exploring the use of open-source LLMs and RAG for enhancing the contextual accuracy of generated responses.

Computer Vision

My work in computer vision has focused on developing robust image classifiers and object detection models using PyTorch and TensorFlow. I’ve tackled challenges such as optimizing Cryo-EM image acquisition, applying transfer learning techniques, and automating image preprocessing workflows across varied data sources, consistently aiming to enhance model performance and efficiency.