My Projects
A collection of projects I've worked on, showcasing different technologies and problem-solving approaches
Featured Projects
NVIDIA-partnered Digital Avatars
Led the co-development of a customized deployment of NVIDIA’s Tokkio avatar platform for healthcare, serving as technical lead for McKinsey. Collaborated closely with NVIDIA engineers to adapt digital human capabilities for real-world clinical use cases.
NVIDIANIMK8sAnsibleMLOps
Commercial Excellence Through Space (COMETS)
Key developer on McKinsey’s COMETS solution, building scalable pipelines to process and infer on satellite and drone imagery. Trained foundational models for remote sensing and developed algorithms that translated geospatial data into actionable commercial insights.
Computer VisionRemote SensingGeospatial AIPythonSatellite Imagery
Other Projects
Salesforce AgentForce Hackathon Winner
Placed second in the Salesforce AgentForce Hackathon, winning $10,000. Built an AI agent solution to optimize the bidding process for SMBs.
Salesforce AgentForceAI AgentsSMB SolutionsProcess Optimization
YesYesNosema - CV Fungal Spore Counter
Developed a machine vision system to automate microscope-based score-counting for Carl Hayden Bee Research Lab, replacing manual labor with CNN-based automation.
OpenCVCNNComputer VisionPythonResearch Automation
GPT-Powered Food Logger
Built a lightweight food logging tool integrating custom GPT with database for macro tracking. Facilitated personal 40-pound weight loss, developed on a flight from Greece.
GPT APIDatabaseNutrition TrackingPythonPersonal Health
Movie Metadata Sync Tool
Built a comprehensive metadata synchronization system that automatically syncs movie collections, ratings, and watch status across multiple platforms including Plex, Letterboxd, and local media applications.
PythonAPI IntegrationData SyncPlex APILetterboxdAutomation
![[Confidential] Agentic Medical Application](/images/genai-infrastructure.png)
[Confidential] Agentic Medical Application
Built a full-stack medical application integrating speech-to-text and agentic AI systems. Developed the React-based frontend and architected cost-optimized backend services using AWS Lambda and Fargate/ECS, all provisioned through infrastructure as code for scalable clinical deployment.
GenAIInfrastructure-as-CodeLLM ServingSpeech-to-TextCloud
Machine Vision Laser Tag App
Created laser tag game using machine vision instead of physical sensors. Used Mask R-CNN to identify players and detect hits, with React/React Native frontend and AWS backend.
Mask R-CNNReactReact NativeAWSComputer Vision
ML Hive Management Tool
Built a predictive tool for the USDA to help beekeepers decide whether to overwinter hives. Trained on field data, the model estimates survival probabilities based on hive health and parasite levels. Accessible through a simple web interface for real-world decision-making.
RNNEnergy OptimizationForecastingPythonThesis Research
DARPA's "Big Mechanism"
Modeled biomolecular reactivity based on functional and structural similarities. Developed hierarchical domain-based similarity prior for MCMC simulations in reaction prediction.
MCMCMolecular ModelingResearchPythonBioinformatics
Honeybee Biology Research
Studied molecular genetics of bee nutrition across 3 peer-reviewed papers. Built logistic regression model for overwintering decisions and analyzed pollen diversity effects on viral titers.
Logistic RegressionMolecular GeneticsResearchStatistical Analysis
AWS Robotics Hackathon
Implemented real-time computer vision on NVIDIA Jetson board for autonomous robot navigation, using machine vision to identify and steer along predefined courses.
NVIDIA JetsonComputer VisionRoboticsReal-time Processing
Autonomous Mine Coordination System
Built system for coordinating self-driving mine equipment using inter-vehicle communication at University of Arizona Hackathon. Received GPS signals and sent optimized routing commands.
GPSInter-vehicle CommunicationRouting OptimizationAutonomous Systems
CHRP-AI
Led undergraduate thesis project optimizing energy usage for University of Arizona campus chillers using RNNs for demand forecasting and energy-efficient pre-activation strategies.
PythonAI/MLMachine LearningData Science