I'm Kozato Keizo.

Studied core computer science principles, including algorithms, data structures, software development, and database management. Gained hands-on experience in full-stack development, cloud computing, and distributed systems through academic projects and research.

Directed comprehensive LoRA fine-tuning pipelines using Azure Foundry, PyTorch, and ONNX, covering data preprocessing, model training/validation, packaging, scalable deployment, and operational monitoring.

Architected and led production Agentic MCP in Python to orchestrate tools, handle API calls, perform secure database operations, and support advanced RAG features (HyDE, Query Decomposition, Weighted Transformation, Text-to-SQL) powering 50k+ daily agent interactions at 200ms P95 latency with 99.99% uptime.

Optimized search performance contributing to an 80% increase in speed through fine-tuning Elasticsearch queries and enhancing search APIs. Orchestrated microservice architecture facilitating distributed APIs and event-driven communication through Kafka, decreasing average API latency by 30%.

Developed robust backend services in FastAPI with scalable structures, integrating Azure SQL Database and MongoDB, Redis-backed multi-level caching. Established Kubernetes platform AKS with Terraform, network policies, and GitOps-style deployment. Delivered full-stack observability with Prometheus + OpenTelemetry + Jaeger.

Technologies: Python, FastAPI, LangChain, LangGraph, PyTorch, ONNX, Azure, Kubernetes, Terraform, Elasticsearch, Kafka, MongoDB, Redis, Prometheus, OpenTelemetry, Jaeger

Implemented ReactJS and Redux to create high-quality, scalable, and reusable components and front-end solutions. Used React.js and Redux framework for premade components from NPM, working with Redux architecture to increase website speed.

Migrated SOAP web service resources to Spring RESTful utilizing Spring REST API and Spring Boot. Created action classes in Struts and developed manager classes using Hibernate, JPA, and Struts technologies, implementing application-level persistence.

Contributed to the company's AI chatbot development, working with the Data team to integrate advanced LLM models, enhancing user interactions and overall functionality. Created and maintained architecture for Restful API utilizing Spring Boot to monitor and manage applications in production.

Developed and executed all phases of testing with Jasmine, Karma for unit testing, and Protractor for End2End automated testing on both client and server sides utilizing Mocha and Chai frameworks.

Technologies: React.js, Redux, Java, Spring Boot, Spring Cloud, Hibernate, JPA, Node.js, MongoDB, MySQL, LLMs, REST APIs, GraphQL

Refined Proton, a command-line tool facilitating faster service creation and updates; drove a 40% increase in developer velocity and was adopted as the most used CLI tool within 6 months.

Innovated Archimedes, a dependency tracker, which identified and helped resolve the three biggest causes of dependency-related build failures, improving build stability and reliability.

Enhanced backend systems with async programming, improving concurrency handling and scalability, increasing overall system performance and stability by 3.5x.

Built advanced React and Angular dashboards in TypeScript to visualize large-scale product and user-flow analytics, delivering smooth performance across 50K+ data points. Engineered an ASP.NET backend powering real-time monitoring tools that visualized service performance, error trends, and dependency health, reducing investigation time by 60%.

Rectified a critical account auto-refill error, collaborating with over 5 external engineers to prevent erroneous card transactions, saving the company $100,000 per month previously lost to reimbursements.

Technologies: C#, ASP.NET, TypeScript, React, Angular, AWS (S3, Lambda, Aurora), Kubernetes, Docker, SaltStack, CI/CD

Discussed, analyzed, and strategized product design with Product Managers, Design Team, and Business Managers. Created and deployed RESTful API endpoints using NodeJs, ASP.NET, and InfoBurst cached XDC queries on AWS.

Developed and designed web application visualizations using JavaScript frameworks such as Angular 5/6/7 and React w/ Redux. Transformed SAP BusinessObjects reports from SAP Web Intelligence into consumable data for API use for multiple client projects.

Created SAP BusinessObjects dashboards proof of concepts for multiple Business Intelligence clients. Crafted dashboards and visualization-heavy applications using UI and Dashboard frameworks such as Telerik, KendoUI.

Formulated multiple POC web applications with integrated OKTA and Auth0 in React and Angular for multiple clients.

Technologies: Node.js, ASP.NET, React, Angular, Redux, AWS, SAP BusinessObjects, Telerik, KendoUI, OKTA, Auth0