Experience
I improved pediatric patient outcomes and reduced malnutrition treatment cost in Gombe, Nigeria.
- Built a facial recognition model with 99% true acceptance/rejection rates and 97% top-three accuracy.
- Forecasted the decay of facial recognition accuracy as malnourished children recover and grow.
- Built and deployed a mobile app for contactless on-device biometric identification.
- Built and automated a backend system to update biometric templates.
- Integrated GCP OAuth 2.0 for secure user authentication in the mobile biometric app.
- Collaborated with field teams and local stakeholders to pilot biometric identification, ensuring ethical and effective implementation in low-resource settings.
- Discovered a cohort of repeat patients and performed analyses to improve treatment protocols for relapse cases.
- Forecasted anthropometric recovery trajectories for malnourished children to identify those most at-risk of nonresponse and death.
- Translated research findings into operational improvements, identifying key implications for malnourished patient protocols and improving long-term recovery.
I lead agile teams delivering scalable AI solutions for precision agriculture — influencing strategic direction, defining quarterly milestones, and aligning research and engineering deliverables with business goals.
- Led cross-functional research teams to develop and deploy AI-based remote sensing, image classification, and LLM frameworks supporting regenerative agriculture.
- Designed advanced AI models for agronomic management practice verification to support regenerative agriculture business operations.
- Deployed and scaled satellite imagery-based predictive pipelines in cloud environments to support high-throughput regenerative agriculture operations.
- Managed the strategic roadmap and delivery of predictive modeling systems, enabling risk assessment tools for farmers and breeders.
- Developed deep learning architectures for joint multi-disease prediction and crop yield forecasting to enable granular, probabilistic risk assessment.
- Engineered scalable production pipelines to process geospatial/environmental data, train deep learning models, and serve disease and yield forecasts.
- Translated complex statistical forecasts and uncertainty metrics into actionable business strategies for non-technical leadership through high-level presentations and reports.
- Authored technical reports and built dashboards that translate performance metrics into intuitive insights and business value.
- Implemented visualization pipelines to securely and reliably deliver predictive forecasts to business decision-makers.
I led agile teams delivering predictive models for crop diseases, defining scientific strategy and aligning research with commercial goals.
- Developed and deployed a deep learning Gaussian process model for jointly forecasting multiple crop diseases.
- Improved data collection efficiency by defining a data valuation strategy and hiring two contractors to execute on it.
- Identified and addressed risks of collecting redundant data across programs.
- Processed large-scale agricultural datasets for model training using Apache Spark for distributed computation.
- Created in-season wheat disease forecasts using probabilistic deep learning models.
- Improved crop yield models by creating deep learning embeddings of high-dimensional environmental data.
- Mentored an intern to develop probabilistic deep learning models forecasting soybean yield over long lead times.
- Identified crop nutrient deficiencies in soil by developing predictive statistical models using satellite imagery.
Reviewed grant proposals for the USDA's Data Science for Food and Agriculture Systems awards.
Skills
Service
Reducing nonresponse and mortality rates of malnourished Nigerian children through predictive model development, deployment, and automation. Assisting with cost-effectiveness analyses.
Advising the University of Florida's Agricultural & Biological Engineering department on mission strategy and curricula to ensure successful graduate placement.
Projects
Donation tracking webapp with database, backend, and frontend deployed on DigitalOcean.
Birthweight prediction intervals using probabilistic deep learning.
Bayesian inference framework for modeling metabolic markers from personal training data.
Developer tools spanning multiple domains and hosted on RapidAPI.
Visualization of publicly available financial data from the Mid-Ohio Food Collective.