Kenneth Leo

Generative AI Video Platform
Generative AI & Agent Architecture

• Architected an end-to-end pipeline to generate authentic UGC-style creator videos, automating persona JSON extraction into canonical facial anchors and 3-angle reference packs with realistic sensor noise and skin texture controls.
• Formulated the "Frame 0" anchoring methodology, enforcing flat smartphone-sensor optics, natural exposure flaws, and strict limb accounting to eliminate facial drift, artificial studio gloss, and physical morphing.
• Implemented 1.0x real-time temporal velocity controls and anti-drift directives to prevent sluggish motion artifacts, while building a multi-provider routing layer across video engines (Alibaba Wan, MiniMax, Seedance) and LLM storyboarding (DeepSeek, Qwen).

Industrial / Mining Company
Industrial Machine Learning & Process Analytics

• Formulated predictive models (XGBoost, multivariate regression) on industrial time-series telemetry to optimize metal recovery in flotation processing circuits, utilizing SHAP values and partial dependence plots to isolate actionable operational levers (chemical dosing, aeration).
• Engineered an early-warning failure framework for heavy crushing operations, isolating discrete stoppage modes (hopper bogging vs. oversized rock jams) using Databricks, plant control system (PCS) logs, and multi-sensor telemetry.
• Developed domain-specific time-series features—including stockpile blend transition detection, lagged grade variables, and recursive feature elimination (RFE)—while enforcing sub-1-hour lookahead constraints to eliminate data leakage under severe class imbalance.
• Built computer vision pipelines (e.g. static ROIs) on surveillance camera feeds to detect haul-truck ore dump cycles and cross-verify fleet dispatch records against physical feed events.

Health and Wellness Company
UI Implementation

• Executed a complete reskin of the brand's Shopify website, translating Figma designs into Liquid templates.
• Overhauled the site’s global styling (CSS/SCSS) to align with the company's new visual brand.
• Refactored parts of legacy theme code to ensure responsiveness and design consistency across mobile and desktop.

Organic Food Company
E-commerce / Web Development

• Collaborated on updating and maintaining the company’s Shopify storefront and brand website
• Customized Shopify Liquid templates and sections to match updated Figma design guidelines and brand aesthetics
• Improved product landing pages, cart drawer components, and customer navigation flow for better conversion and usability

Internal Project
Backend / Frontend Development

• Fixed existing bugs and optimized data flow in Django backend to improve API reliability and performance
• Implemented new functionalities in the React.js frontend to enhance user experience and streamline internal workflows
• Collaborated with team members to test and deploy updates, ensuring smooth integration between backend and frontend systems

Machine Learning Training & Applied Projects (Kaggle & Internal Exercises)
Model Development / MLOps

• Built and evaluated predictive models on real-world tabular datasets using PyTorch Lightning and XGBoost, following Kaggle-style competition workflows.
• Implemented end-to-end training pipelines, including data preprocessing, feature engineering, training, validation, and inference.
• Used MLflow for experiment tracking, metrics comparison, and model versioning across multiple runs.
• Orchestrated reproducible workflows with Prefect, and performed automated hyperparameter tuning using Optuna and Ray.
• Applied best practices in cross-validation, leakage prevention, and performance analysis to improve model robustness.

• Designed and delivered structured lectures and laboratory courses in Classical Mechanics, Electromagnetism, and Optics, with a strong emphasis on analytical and quantitative problem-solving.
• Taught computational and data-driven methods including algorithms, numerical modeling, machine learning fundamentals, image processing, and scientific computing using Python and MATLAB.
• Developed course materials and hands-on projects involving data analysis, simulation, and modeling, reinforcing practical programming and analytical skills.
• Guided and evaluated student technical projects and research outputs for academic competitions and conferences, applying rigorous code review, model validation, and results interpretation.
• Strengthened expertise in statistical reasoning, data pipelines, and reproducible analysis, directly applicable to data science, machine learning, and software engineering workflows.