Christopher M. Stewart, Ph.D.
AI / ML Research & Development
Researcher | Data Scientist | ML Engineer
AI researcher and applied scientist with 15+ years bridging academic research, health informatics, and large-scale industry AI/ML. Experience includes 8.5 years at Google deploying LLM-powered content safety systems across 10+ production launches. Currently leading multi-institutional research on applied AI/ML with collaborators at Carnegie Mellon, Google, St. Jude Children’s Research Hospital, among others.
Competitive Edge: Rare combination of hands-on production ML engineering (300K+ lines of production code at Google), original research published in top venues, and deep theoretical grounding in computational linguistics and statistical modeling with experience translating cutting-edge research into production-ready implementations at scale.
Core Competencies:
Highlights
Production ML at Google scale: Drove end-to-end delivery of LLM-powered content-safety systems across 10+ high-impact production launches, preventing >$100M of policy-violating monetization. Added 300K+ lines of production code.
Healthcare AI & safety research: Co-leading a safe-AI research project on medical misinformation and health disparities in cancer survivorship. Delivered medical misinformation classifiers at the height of the COVID-19 pandemic at Google. Published a systematic review on NLP for patient-reported outcomes in Artificial Intelligence in Medicine.
Technical depth & impact: Conducted original research on multi-modal many-shot prompting and graph-based LLM inference. Leading research comparing Bayesian and neural retrieval methods for RAG. Won multiple hackathon awards at Google and published a defensive patent.
Experience
Leading and collaborating on multiple original research projects spanning AI safety, health informatics, and retrieval-augmented generation with researchers at Carnegie Mellon University, Google, St. Jude Children’s Research Hospital, Aarhus University, and the University of Zürich.
- Leading research comparing Bayesian and neural retrieval methods for RAG
- Leading research on making AI benchmarking robust to changes in ground truth via psychometric methodologies
- Co-leading a safe-AI research project on medical misinformation and health disparities in cancer survivorship
- Co-leading research on AI-generated health language alignment
- Submitted book manuscript Building Language Technologies (under contract, MIT Press)
- Completing Stanford School of Engineering’s AI Professional Program
- Anticipated output: 7+ peer-reviewed publications, 1 book, and Stanford AI Professional Certificate
Senior Computational Linguist / Research Scientist (L5), 03/2022–08/2025
- Drove end-to-end delivery of LLM-powered content-safety systems across 10+ high-impact production launches, preventing >$100M of policy-violating monetization
- Delivered medical misinformation classifiers at the height of the COVID-19 pandemic
- Designed novel detection methodologies for spam, fraud, and abuse that were presented to Google’s C-suite
- Conducted original research on multi-modal many-shot prompting and graph-based LLM inference
- Coordinated cross-functional programs spanning ML engineering, policy, product, and global rater operations
- Added 300K+ lines of production code, won multiple hackathon awards, and published a defensive patent
Analytical / Computational Linguist (L3 → L4), 03/2017–03/2022
- Proposed, designed, and built automated statistical analysis tooling for large-scale crowd-sourced data collection, adopted by 100+ global teams
- Defined requirements and evaluation workflows for a worldwide rater workforce producing training and validation data for production models
- Evaluated rater reliability at scale using ML and psychometric methods
Senior Data Scientist
- Built analytics pipelines and forecasting models for high-frequency IoT sensor data as sole Data Scientist at an Industrial IoT startup
- Developed supervised learning package in R
- Mined large-scale socio-demographic data for Walmart, including novel propensity-score methods
Voice Engineer
- Engineered synthetic voices for Apple’s Siri using large-scale multimodal speech data within a distributed, multi-national R&D team
- Led customer-facing workstream for reboot of the world’s most widely heard synthetic voice
Assistant Professor
- Initiated multidisciplinary research resulting in 5 publications, 5 conference papers, and 2 invited talks
- Co-PI on a neuroscience grant
- Revamped course sequence, raising enrollment 86%
- Supervised team of 10 instructors
Leadership & Mentoring
- Co-organized Linguistics Career Launch, mentoring graduate students at American and European universities transitioning into industry careers in NLP and computational linguistics.
- Invited talks at 13+ institutions including Google Faculty in Residence Program, Georgetown University, University of Wisconsin-Madison, LSA, and University College London.
Selected Publications (full bibliography on Google Scholar)
Forthcoming
Li, M., Tang, N., Stewart, C.M., Heidari, H. & Shen, H. (2026). “Between Rigor and Reality: How AI Safety Benchmark Developers Understand Benchmark Use and Maintenance.” (Submitted to ACL 2026 EvalEval Workshop)
Bohojlo, C. & Stewart, C.M. (2026). “Policy Over Prompts: RL-Based Selection for Financial Question Answering.” (To be submitted to EMNLP 2026)
Stewart, C.M. et al. (2026). “Addressing Medical Misinformation and Health Disparities in Cancer Survivorship Using Safe AI.” (To be submitted to BioNLP Workshop 2026)
Stewart, C.M., Kraus, K., Brown, G., & Casal, J.E. (2026). “Who’s Talking? LLM Age Inference Across Text, Speech, and Multimodal Input.” (To be submitted to Ninth AAAI/ACM Conference on AI, Ethics and Society)
In Press
Stewart, C.M., Tyler, J. & Thyme-Gobbel, A. (2026). Building Language Technologies. MIT Press.
Published
Casal, J.E., Stewart, C.M., & Windsor, A.J. (2025). “‘It Is Important to Consult’ a Linguist: Verb-Argument Constructions in ChatGPT and Human Experts’ Medical and Financial Advice.” PLoS One 20(5): e0324611.
Sim, J.A., Huang, X., Horan, M.R., Stewart, C.M., et al. (2023). “Natural Language Processing with ML Methods to Analyze Unstructured Patient-Reported Outcomes from EHRs: A Systematic Review.” Artificial Intelligence in Medicine 146: 102701.
Education and Professional Development
University of Illinois at Urbana-Champaign
- Awarded 8 fellowships. Published 3 articles. 5 conference presentations.
Furman University
- 2026Stanford School of Engineering Artificial Intelligence Professional Program
- 2016Coursera Machine Learning Specialization
- 2015Coursera Data Science Specialization
Technical Skills / Natural Languages
- Languages
- Python, R, SQL, Bash, HTML/CSS/XML
- AI/ML & NLP
- PyTorch, TensorFlow/TFX, LLM development via API (Gemini, GPT-4/5, Claude, Mistral), supervised fine-tuning, RLHF & preference optimization, prompt engineering, adversarial red-teaming, agentic AI, RAG, embedding models & vector search, multimodal integration, speech feature extraction, supervised / unsupervised / reinforcement learning
- Methods
- Mixed-effects models, causal inference, Bayesian modeling, predictive modeling, sentiment analysis, topic modeling, feature engineering for high-dimensional data
- Infrastructure
- Distributed computing, Hive, Spark, Git/GitHub, Jupyter. Google-internal: Critique, Subversion
- Natural Languages
- English (native), French (near-native), Spanish (highly proficient), Serbo-Croatian & Italian (conversational), German & Dutch (elementary)
Recommendations
Recommendations from managers and co-workers can be found on my LinkedIn profile.
Other references available upon request.