I’m a fourth-year BS Computer Science student at the University of Cabuyao pursuing a career in software quality assurance, with experience in web development, requirements-based workflow checks, and technical project coordination.
Through student leadership and community partnerships, I’ve developed skills in team coordination, organization, and clear communication.
I’m seeking an entry-level opportunity to strengthen my software testing skills and contribute to reliable, user-focused software.
Tech Stack
01 Web Development
HTML5CSS3JavaScriptReactNext.jsAstroTailwind CSS
02 Languages & Core
JavaPythonPHPTypeScript
03 Backend & Platforms
Node.jsMySQLFirebaseSupabaseJSONVercelRender
04 Tools & Design
GitGitHubDockerFigmaCanvaCodexClaude
Experience [10]
01
Association of Computer Science Students (ACSS) — UC (PnC)
President
2024 – Present
Public Relations Officer
2023 – 2024
02
No-Code Philippines
Director for Partnerships and Business Development
2026 – Present
Community and Events Committee Member
2025
03
DEVCON Laguna
Partnership Liaison Officer
2026 – Present
Speaker Relations Committee Volunteer
2024
04
PROJECT XP: Experience Digital Growth in Laguna's Arcade of Learning
Founder
2025 – Present
05
n8n Philippines User Group
Workflow Department Member
2026 – Present
06
College of Computing Studies – Student Government
4th Year Representative
2026 – Present
Scout
2023 – 2024
07
PnC Lente
Layout Artist
2025 – Present
Photographer
2024 – 2025
08
Devkada
Partnerships Officer
2026
09
Quantum Computing Society of the Philippines
QE Fellow, Operations and Talent Development Committee
2025
10
ASEAN Youth Advocates Network — Global
Public Relations Officer
2024 – 2025
External Volunteer for Creatives
2024
04
Projects
01
HorizonAIThesis / AI Refactoring
PythonFastAPINext.jsllama-cpp-pythonSQLite
Our bachelor’s thesis: an offline multi-agent system for Java refactoring within 4 GB of VRAM. HorizonAI coordinates planning, code generation, and validation through specialized small language models, evaluated on 279 Java tasks to examine memory efficiency and refactoring reliability.
Our entry submitted for the Philippine Startup Challenge 10: a community platform concept connecting veterinary clinics, rescue groups, and animal-care supporters. Its landing page introduces the vision, outlines planned features, and gathers early interest through role-based registration and a public roadmap.
A client landing-page project for Anytime Fitness, built with React, TypeScript, and Tailwind CSS. The page brings club discovery, membership information, and coaching into one experience, with a preview available while official utilization is still pending.
Project XP Arcade ToolkitCommunity / Arcade ToolkitUnder development
ReactViteReact RouterTailwind CSS
An arcade-style digital literacy toolkit for Grades 4–6 learners, developed as part of PROJECT XP, the 2025 SIBOL ng Laguna Kaunlarang Teknolohikal champion initiative. Still under development, it brings learning games, facilitator resources, and community documentation into one platform.
Our second-year OOP final project: a Java Swing learning management system for students, instructors, and administrators. NovaLearn connects courses, learning content, notifications, and performance tracking through object-oriented design, using Singleton, Factory, and Observer patterns with in-memory object lists.
Highlights [5]
Credentials [3]
Agentic AI Certified Foundations Associate
Oracle / 10.31.2026
I earned Oracle’s foundations certification in agentic AI, covering agent architecture, reasoning, tool use, multi-agent workflows, and safety guardrails. This strengthens my technical understanding of AI systems and gives me a clearer basis for discussing their behavior, limitations, and validation needs.
I earned Google’s Gemini Certified Student credential, covering effective prompting, response evaluation, and responsible AI use for learning and productivity. It strengthened my approach to AI-assisted work: reviewing generated outputs critically and using these tools thoughtfully when researching, communicating, and organizing tasks.
I completed 42 hours of entrepreneurship training and developed a Certified Practice Venture through Wadhwani Foundation’s Ignite Philippines program. The experience strengthened my problem discovery, idea validation, prototyping, and pitching skills, helping me connect technical project decisions with user needs and practical delivery.
I'm always interested in meaningful software projects, open-source initiatives, and collaborative ventures. Let's connect and build something impactful together.
psst, say hi
[ open to opportunities ]Replies within 24 hoursCabuyao, Laguna · GMT+8
PROJECT / 01
HorizonAI
Thesis / AI Refactoring
Our bachelor’s thesis: an offline multi-agent system for Java refactoring within 4 GB of VRAM. HorizonAI coordinates planning, code generation, and validation through specialized small language models, evaluated on 279 Java tasks to examine memory efficiency and refactoring reliability.
My role
Overall Project LeadKept development, evaluation, and documentation aligned across the thesis team.
Enabled fully offline Java refactoring within 4 GB VRAM, with 20.1% lower peak memory usage than the baseline.
01 / CONTEXT
The project goal
Make controlled, offline Java refactoring accessible on a consumer laptop with only 4 GB of VRAM. Our Bachelor of Science in Computer Science thesis at the University of Cabuyao investigated whether specialized small language models could coordinate within that limit, with explicit validation at every handoff.
02 / APPROACH
What shapes the experience
Strict Sequential Inference Scheduling (SSIS) formalizes Sequential Model Loading: only one model executes at a time, keeping the multi-agent workflow within the memory budget.
The Planner and Generator share 4-bit Qwen2.5-Coder-3B-Instruct weights. Distinct prompts, output schemas, and a context reset separate their roles without an extra weight swap. The Judge uses 4-bit Llama-3.2-3B-Instruct.
AST-Constrained Semantic Alignment maps the requested change to code boundaries. Before judging, javalang syntax parsing and structural checks screen generated Java.
Dual-Loop Deterministic-Neural Routing (D2NR) combines deterministic validation with neural review. Failed candidates return for bounded correction; retry limits contain failures and support recovery or restoration of the original code.
A local Glassbox interface exposes the workflow, agent handoffs, output, and logs so users can inspect the process rather than receive an unexplained code replacement.
03 / ABSTRACT
Small models, explicit governance
Larger language models demand computing resources many schools and developers lack. A single model also handles every reasoning step, allowing an early error to propagate into generated code. HorizonAI investigates horizontal intelligence: a planner, generator, and judge coordinated through memory scheduling and validation.
Using Java refactoring as its application domain, the study compared HorizonAI with a larger single model on 279 tasks and gathered feedback from 25 end users and 10 IT experts. The architecture met the 4 GB VRAM constraint and resolved 78.5% of tasks. It used 20.1% less peak memory and produced slightly simpler code, but took about five times longer. Its compilation and behavior-preservation rates reached 86% and 88% of the baseline rates, respectively.
The contribution is a measured tradeoff: coordinated small models can support controlled local refactoring under a strict hardware ceiling. The results establish lower memory use and offline accessibility, without claiming higher accuracy than the larger baseline.
04 / RESEARCH PROBLEM
Why resource constraints shaped the system
Vertical model scaling raises hardware and energy requirements. For institutions without high-end GPUs, local execution becomes impractical, while cloud inference introduces recurring costs and sends code outside the local environment.
Reliability is a separate challenge. Probabilistic code generation can introduce defects or unnecessary complexity, and a reasoning error can cascade through a monolithic workflow. The study addresses the limited empirical evidence for structured, fully offline multi-agent refactoring under severe hardware constraints.
Conducted at the University of Cabuyao from January to October 2026, the work deliberately reflects resource limitations faced by nearby institutions such as St. Vincent College of Cabuyao and Westbridge Institute of Technology, Inc.
05 / ARCHITECTURE
Five modules support one local workflow
Orchestration and model execution
The Orchestrator governs stage transitions and commands the Agent Service through SSIS. The Agent Service loads and executes quantized models sequentially. Sharing the Planner and Generator weights reduces model swaps while preserving separate responsibilities.
Validation and correction
The Validator applies syntax and structural checks before the Judge reviews the result. These deterministic gates and model-based revision requests form the D2NR correction loops; neither replaces a formal proof of behavioral correctness.
Session context and interface connection
The Context Manager stores session state, reasoning logs, and agent handoffs in local SQLite so information survives model unloading. The Connection Manager links the offline backend with the Glassbox web interface for real-time monitoring.
Implementation stack
The repository implements a Python/FastAPI backend with llama-cpp-python and a Next.js/TypeScript frontend. WebSocket updates connect the interface to the local pipeline; Docker distribution supports local installation. Repository capabilities may evolve beyond the evaluated thesis configuration.
06 / EVALUATION
A quantitative comparison on 279 Java tasks
Hardware and baseline
The execution environment was an Intel Core i5 laptop with 16 GB RAM and an NVIDIA RTX 2050 with 4 GB VRAM. All evaluated models used 4-bit quantization. A Qwen2.5-Coder-7B-Instruct model provided the larger single-model baseline.
Benchmark and metrics
The benchmark used 279 Java tasks from the Code Polish subset of CodeEditorBench. Compilation Success Rate measured whether generated code compiled; Behavioral Equivalence Rate measured preservation of original behavior. Local inference time and peak GPU memory captured the hardware cost, alongside code-complexity observations.
User and expert evaluation
Twenty-five purposively sampled computing students and faculty evaluated interaction capability, functional suitability, security, and reliability. Ten IT experts assessed functional suitability, performance efficiency, reliability, and maintainability using the ISO/IEC 25010:2023 product quality framework.
07 / RESULTS
Lower memory use with measurable costs
Task resolution and memory
HorizonAI stayed within the 4 GB VRAM limit and resolved 78.5% of the selected tasks. Peak memory use was 20.1% lower than the larger single-model baseline, and the resulting code was slightly simpler.
Runtime and relative correctness
Execution took approximately five times as long. Compilation success reached 86% of the baseline’s compilation rate, while behavior preservation reached 88% of the baseline’s behavior-preservation rate. These are relative comparisons, not absolute success percentages or percentage-point differences.
User acceptance and expert assessment
End users rated the system highly acceptable, and technical experts rated all four assessed quality characteristics favorably. The evaluation covered interaction capability, functional suitability, security, reliability, performance efficiency, and maintainability across the two participant groups.
08 / SCOPE & LIMITS
What the evidence supports
Supported workload
The evaluated workflow processes text and isolated Java snippets. It does not cover complex multi-file repositories, multimodal input, model fine-tuning, or distributed deployment. Unsupported transformations, including Extract Class, fall outside the implemented benchmark coverage.
Offline execution and latency
Inference runs fully offline once the required software and quantized weights are available locally. Reasoning capacity is bounded by those local weights. Sequential execution conserves memory but increases latency and limits responsiveness.
Validation boundaries
Syntax parsing and structural routing detect specific failures; they cannot mathematically guarantee alignment with developer intent or perfect behavior preservation. Findings are limited to the selected tasks, purposive survey samples, quantized models, and tested laptop configuration.
Screenshot context
The screenshots document the local interface, including guided-tour examples and an individual session’s metrics. They are interface illustrations, not aggregate benchmark evidence.
09 / SIGNIFICANCE
A reference for accessible local AI
AI researchers and system designers
For AI researchers and system designers, SSIS and D2NR provide documented patterns for memory-constrained orchestration, bounded retries, and recovery. Future work can investigate other languages, tasks, models, or hardware limits.
Practitioners and local development teams
For software engineering practitioners, the visible stages and validation gates support inspection before integration. Philippine development teams and MSMEs can use the findings to assess offline deployment, proprietary-code privacy, hardware affordability, and recurring API costs.
Academic institutions
For academic institutions, the tested consumer laptop provides a concrete reference for teaching multi-agent coordination without a specialized AI laboratory. Applicability depends on comparable hardware and workload rather than a universal performance guarantee.
10 / TEAM
The thesis team
Christian BalinadoMe
Overall Project Lead
Overall project direction, team coordination, and oversight of thesis deliverables.
Joshua Lopez
Lead / Fullstack Developer
Technical implementation across the frontend and backend, including integration of the refactoring workflow.
Jericho Varde
Frontend Developer
Implementation of the Glassbox web interface and its user-facing refactoring views.
Andrew Dejito
Quality Assurance Manager
Quality assurance coordination, testing oversight, and verification of system behavior against project requirements.
11 / SCREENS
A look at the project
Active refactoring session with the Planner stage in the simple flow viewHorizonAI welcome screen for the local multi-agent refactoring applicationGuided tour: Java code editor and refactoring instruction inputGuided tour: detailed pipeline from baseline validation to finalizationGuided tour: live terminal with agent decisions and validation logsIndividual session insights showing complexity, inference time, GPU use, and memory
1 / 6
12 / DELIVERABLE
What was built
A locally deployed, role-driven Java refactoring system and an empirical evaluation of its memory, runtime, correctness, and software quality. The thesis demonstrates feasibility on modest hardware while documenting the cost in speed and accuracy. The public landing page introduces the project; refactoring runs in the separately installed local application.
Our entry submitted for the Philippine Startup Challenge 10: a community platform concept connecting veterinary clinics, rescue groups, and animal-care supporters. Its landing page introduces the vision, outlines planned features, and gathers early interest through role-based registration and a public roadmap.
My role
Team LeadHelped the team translate the platform concept into a focused pre-launch page.
Tech stack
React / Next.js / CSS / Vercel
Key outcome
Delivered a public landing page with role-based early-access registration and a product roadmap.
01 / CONTEXT
The project goal
Introduce Animora, our Philippine Startup Challenge 10 entry, through a focused pre-launch landing page. The concept connects veterinary providers, rescue groups, and animal-care supporters in the Philippines. The page gives visitors a clear introduction to that vision and a way to express early interest before the wider platform is built.
02 / APPROACH
What shapes the experience
Lead with a simple community-care message and a prominent early-access form instead of presenting the entire platform outline at once.
Separate interest in sponsoring animal care from interest in offering veterinary services through role selection at registration.
Group the proposed benefits into scannable sections covering trust, rescue support, and community participation.
Give the roadmap its own destination so visitors can distinguish the landing-page release from later platform milestones.
03 / AUDIENCE
One concept, two clear ways to participate
The outline describes a fragmented animal-care ecosystem: pet owners need accessible care, rescue groups need support and visibility, and veterinary providers need ways to connect with their communities. The landing page condenses that broad problem into an invitation to care, connect, and contribute.
The opening form asks visitors whether they want to sponsor animal care or offer veterinary services, followed by an email field and an early-access action. This makes the intended audience visible at the point of registration without requiring visitors to understand every planned feature.
Animora introduction with two participation roles and an early-access email form
04 / PAGE DESIGN
Make a broad platform idea easy to scan
A warm yellow-to-orange palette, generous spacing, and a short headline give the page a welcoming tone. Home and Roadmap navigation keep the main introduction and future direction easy to find.
The feature comparison and benefit cards organize the proposed value into short explanations. Rather than reproducing the full business outline, the page emphasizes the themes most relevant to prospective participants: local animal care, verified partners, transparent support, and community updates.
Feature comparison presenting Animora’s proposed positioningBenefit cards explaining the proposed community animal-care experience
05 / ROADMAP
Show what comes after the landing page
A separate timeline presents the landing-page release and subsequent plans for partner onboarding and platform development. It gives interested visitors context for what they are joining and how the concept is intended to progress.
The roadmap is a communication artifact from the pre-launch stage. Its dates and later milestones describe the plan shown on the page, rather than verified completion of the wider product.
Animora roadmap separating the landing-page launch from later product milestones
06 / SCOPE & LEARNING
A pre-launch foundation for the startup concept
My contribution
My role was Team Lead for the entry. The focus of this deliverable was helping the team turn a broad platform proposal into a clear public introduction, with an immediate registration action and a visible product direction.
Delivered scope
Veterinary bookings, payments, donation tracking, partner verification, rewards, and clinic tools belong to the proposed platform. Their appearance in the page’s marketing content is not evidence that these services were implemented in the landing-page release.
Evidence and next steps
The screenshots document the page and its registration interface. They do not establish verified signup totals, partnerships, or animal-welfare outcomes. A next step would be to evaluate registration completion and learn which participant groups show interest before expanding the product.
07 / SCREENS
A look at the project
Animora landing page and early-access registrationAnimora platform feature comparisonAnimora community animal-care feature overviewAnimora product roadmap
1 / 4
08 / DELIVERABLE
What was built
A public React and Next.js landing page deployed on Vercel, with an early-access registration interface, a feature overview, and a roadmap. As Team Lead, I helped the team translate the startup concept into this pre-launch deliverable. The result is a place to introduce Animora and invite interest; it does not establish that the proposed animal-care services are operational.
A client landing-page project for Anytime Fitness, built with React, TypeScript, and Tailwind CSS. The page brings club discovery, membership information, and coaching into one experience, with a preview available while official utilization is still pending.
My role
Landing Page DesignerOrganized club discovery, membership, and coaching into a clear visitor journey.
Delivered a landing-page design covering club discovery, membership, and coaching; official utilization is pending.
01 / CONTEXT
The project goal
Design a landing page for Anytime Fitness that introduces the fitness experience and gives prospective members clear ways to find a club and explore membership. This was my first paid project, marking the move from academic work to a commissioned design deliverable. The page is available as a preview while official utilization remains pending.
02 / APPROACH
What shapes the experience
Build a clear visual hierarchy around 24/7 access, club discovery, membership, and coaching, with prominent calls to action throughout the page.
Use a purple-and-aqua visual palette, a distinctive clock-inspired hero graphic, and alternating light and dark sections to organize the browsing experience.
Support exploration with searchable club cards, city filters, club-detail views, membership information, and a coach showcase.
Implement the page with React and TypeScript, Vite, Tailwind CSS, Framer Motion, and Lucide React icons; provide a Vercel preview for review.
03 / DESIGN DIRECTION
Make the first impression clear and actionable
The opening section pairs a large typographic message with a clock-inspired 24/7 visual. Purple and aqua accents connect the brand presentation to the main actions: finding a gym and exploring membership.
A compact navigation bar gives visitors direct routes to the club finder, experience, joining information, coaches, and contact section. The hierarchy keeps those destinations visible without making the introduction compete with every detail below it.
Anytime Fitness landing-page hero with 24/7 graphic and gym and membership actions
04 / CLUB DISCOVERY
Turn interest into a place to start
The club finder combines a search field with city filters and location cards. Each card groups a club name, location, amenities, and a details action, helping visitors compare options in a consistent format.
The implementation filters a local location dataset by city, club name, and area, then opens the selected club in a detail modal. This brings useful interaction to the landing-page design without presenting the preview as a live club-directory integration.
Club finder with city filters, search, amenity tags, and location cards
05 / EXPERIENCE & MEMBERSHIP
Give the benefits a readable structure
The experience section shifts into a dark visual chapter. A consistent set of cards organizes the benefits into short, scannable explanations, with icons and contrasting headings providing clear entry points.
Membership information uses a two-column composition: an introduction and action on one side, with an included-benefits panel on the other. This separates the main message from supporting detail while keeping the next step nearby.
Dark experience section with fitness-benefit cardsMembership section with a benefits panel and membership-options action
06 / COACHES & IMPLEMENTATION
Keep the presentation consistent across sections
Coach presentation
Coach cards use large photography, role labels, and names to introduce the coaching offer. The repository separates coach data access from the React components, with a mock provider as the default and a separate HRIS provider option.
Reusable frontend structure
React components divide the page into focused sections. TypeScript describes shared data, Tailwind CSS supports responsive layouts, Framer Motion supports animated presentation, and shared theme tokens coordinate the light and dark chapters.
My contribution and delivery status
As the landing-page designer, I shaped the visual direction, content hierarchy, and presentation of the visitor journey. The design brings club discovery, membership, and coaching into one consistent experience. The Vercel page remains available for viewing while its official utilization is pending.
Coaching showcase with photographic trainer cards and role labels
07 / SCREENS
A look at the project
Anytime Fitness hero with 24/7 visual and calls to actionSearchable club finder and city filtersFitness experience and benefits cardsMembership overview and included benefitsCoach showcase with photographic cards
1 / 5
08 / DELIVERABLE
What was built
A client landing-page design with an accessible preview, covering the hero, club finder, fitness experience, membership, coaching, and contact sections. My role focused on the landing-page design and the presentation of its content and interactions. Official utilization is still pending; the preview represents the delivered design rather than a confirmed official launch.
An arcade-style digital literacy toolkit for Grades 4–6 learners, developed as part of PROJECT XP, the 2025 SIBOL ng Laguna Kaunlarang Teknolohikal champion initiative. Still under development, it brings learning games, facilitator resources, and community documentation into one platform.
My role
Project Lead / Strategic DirectionConnected toolkit priorities with the wider initiative’s community learning goals.
Tech stack
React / Vite / React Router / Tailwind CSS
Key outcome
Created an arcade-style digital-literacy platform for Grades 4–6 as part of PROJECT XP; toolkit development is ongoing.
01 / CONTEXT
The project goal
Make digital literacy engaging and practical for Grades 4–6 learners in Cabuyao’s underserved communities. The Arcade Toolkit is the digital component of PROJECT XP: Experience Digital Growth in Laguna’s Arcade of Learning, a wider youth-led initiative by the Association of Computer Science Students at the University of Cabuyao. The toolkit remains under development.
02 / APPROACH
What shapes the experience
Translate digital-literacy topics into short, arcade-inspired challenges that mentors can use alongside guided discussion and hands-on activities.
Connect games, facilitator resources, a community gallery, and project information in one recognizable learning platform.
Support the wider initiative’s school and barangay sessions with materials intended for reuse by educators, youth volunteers, and local partners.
Keep the toolkit’s development status separate from the recognition earned by the broader PROJECT XP initiative.
03 / INITIATIVE & RECOGNITION
One toolkit within a wider champion initiative
PROJECT XP was submitted by the Association of Computer Science Students, University of Cabuyao, for Kaunlarang Teknolohikal under Gawad Kampeon para sa Kabataan ng Laguna — “SIBOL,” Kabataan para sa Magandang Laguna.
The wider PROJECT XP initiative became the first-ever champion of the Kaunlarang Teknolohikal category at SIBOL ng Laguna 2025. That recognition belongs to the whole initiative: its community learning model, youth leadership, and supporting activities. The Arcade Toolkit featured here is one component and is still under development.
The full initiative combines mentor preparation, pop-up learning arcades, hands-on activities, and support for parents and teachers. The platform gives that work a digital home and a foundation for making the learning experience reusable.
04 / TOOLKIT PLATFORM
A welcoming entry into digital learning
The home screen uses arcade imagery, pixel-style headings, and a warm yellow palette to frame learning as an adventure. Clear Play Games and Explore Toolkit actions give learners and facilitators different entry points.
Navigation separates Games, Gallery, Toolkit, and About. This structure connects the learning activities with their educational purpose, community setting, and the people behind the initiative. The current platform is a development preview, with further toolkit work still in progress.
Project XP Arcade Toolkit home screen with Play Games and Explore Toolkit actions
05 / LEARNING EXPERIENCES
Four themes, familiar community situations
Cybersecurity and privacy
Cyber Tanod: Barangay Defense Force frames password safety and data protection around defending a barangay. It translates security concepts into a familiar local setting for young learners.
Online safety and digital citizenship
Cyber Tanod: Digital Tambayan introduces scam awareness, suspicious messages, and phishing red flags. Guided activities are intended to help pupils connect game choices to safer behavior online.
Misinformation awareness
Truth Tracker: Live Newsroom focuses on checking sources and separating facts from misleading information. It reflects the wider initiative’s fact-checking and misinformation learning strand.
Productivity and digital habits
Cyber Tanod: Workflow Warrior introduces productive technology use and healthy digital habits. The broader learning plan includes file handling, documents, collaboration, and email practice, supported by youth mentors.
Arcade game catalogue featuring Barangay Defense Force, Digital Tambayan, Truth Tracker, and Workflow Warrior
06 / COMMUNITY LEARNING
Designed for learners and the adults who guide them
Grades 4–6 pupils are the primary beneficiaries. Parents and teachers help reinforce responsible digital habits outside the sessions, while barangay councils, Sangguniang Kabataan groups, and local partners support access and continuity.
The platform’s About page explains the initiative’s educational, youth-led, and not-for-profit character. Its impact gallery provides a place to document school visits and community activities, with filters for the locations shown in the preview.
The toolkit is intended to support facilitated learning rather than leave children to interpret every activity alone. Youth mentors connect the games to conversations about privacy, online behavior, checking information, and everyday technology use.
Project XP goals and beneficiaries, including Grades 4–6 learners and community facilitatorsImpact gallery presenting school visits and community activities
07 / DEVELOPMENT & GOALS
Build for reuse and evaluate the learning
Development status
The Arcade Toolkit remains under development. The current preview demonstrates its visual direction, game catalogue, gallery, and project information; the complete reusable toolkit is an ongoing deliverable.
Planned community reach
The initiative’s original plan targeted 30–40 youth mentors, four barangay pop-up arcades reaching 280–320 pupils, and around 100 parents and teachers through supporting clinics. These are program targets, not measured outcomes attributed to this platform.
Learning evaluation
The plan set a target of at least 30% improvement in digital-literacy test scores. Assessment rubrics and guided activities are intended to make learning progress observable; that target is not presented as an achieved result.
Long-term reuse
The intended open-source toolkit brings together lesson cards, game scripts, and assessment rubrics for local adaptation by CESD, barangays, and SK councils. The wider proposal aligns with SDGs 4, 16, and 17 through education, informed participation, and partnerships.
08 / PROJECT TEAM
Leadership and shared responsibilities
The Association of Computer Science Students at the University of Cabuyao leads the wider initiative. These responsibilities describe the project proponents’ work across PROJECT XP, including the direction and support needed for the Arcade Toolkit.
Project XP proponents and implementing organization
Christian BalinadoMe
Project Lead / Strategic Direction
Overseeing the overall execution and direction of the initiative.
Jenny Losamia
OIC-Secretary / Communications
Handling official communications and project documentation.
Rachel Joy Ramos
Treasurer / Finance
Managing project funds and financial allocations.
Correllene Ira Salar
Auditor / Quality Control
Ensuring transparency and quality in all project outputs.
Gabriel Diamante
OIC-PRO / Public Relations
Managing external relations and community engagement.
09 / SCREENS
A look at the project
Project XP Arcade Toolkit home screenFour arcade-inspired digital-literacy learning themesProject XP community and school activity galleryLearning goals and beneficiariesProject XP proponents and their responsibilities
1 / 5
10 / DELIVERABLE
What was built
An Arcade Toolkit platform under active development, with a game catalogue, project information, and community gallery shown in the current preview. It supports the wider PROJECT XP initiative rather than representing the entire program. My role is Project Lead for the initiative, overseeing its overall execution and strategic direction.
Our second-year OOP final project: a Java Swing learning management system for students, instructors, and administrators. NovaLearn connects courses, learning content, notifications, and performance tracking through object-oriented design, using Singleton, Factory, and Observer patterns with in-memory object lists.
My role
Overall Project Lead & Quality AssuranceChecked student, instructor, and admin workflows against course requirements.
Tech stack
Java / Swing
Key outcome
Built a Java Swing LMS connecting student, instructor, and admin workflows for our OOP final project.
01 / CONTEXT
The project goal
Build a Java desktop learning management system that connects student learning, instructor workflows, and account administration. NovaLearn was our second-year Object-Oriented Programming final project: an opportunity to turn classes, relationships, and design patterns into a usable multi-form Swing application.
02 / APPROACH
What shapes the experience
Organize the application around three roles: Student, Instructor, and Admin, with a shared visual language and role-specific navigation.
Model accounts, courses, learning content, activities, grades, and notifications as collaborating objects rather than a collection of unrelated screens.
Keep demonstration records in List-based object collections, meeting the course constraint of using no database.
Connect the interface to dedicated managers, a content factory, notification listeners, and report services, as documented in the UML design.
03 / ROLE-BASED ACCESS
One application, three responsibilities
The entry screen separates Student, Instructor, and Admin login paths. This makes the available workflows clear before users enter the application and supports distinct responsibilities throughout the design.
The UML places shared account information in User, with Student, Instructor, and Admin representing specialized roles. AccountManager centralizes account collections, account selection, and logout, giving the different forms a shared view of the active session.
NovaLearn entry screen with Student, Instructor, and Admin login options
04 / STUDENT EXPERIENCE
Course content and a personal learning workspace
The student interface brings Courses, Grades, Notifications, and Account into a consistent sidebar. The course stream shows dated learning materials and completion indicators, while Stream and Coursework tabs separate learning content from activity work.
Profile management gives students a dedicated account view with update-profile and change-password actions. In the object model, Student owns profile data and keeps collections for enrolled courses, grades, and notifications. Course enrollment and activity submission are modeled as domain actions rather than database operations.
Student course stream with dated Java learning materials and completion indicatorsStudent account view with profile details, Update Profile, and Change Password actions
05 / INSTRUCTOR WORKFLOWS
From a teaching overview to individual student records
The instructor dashboard summarizes assigned courses, total students, and new submissions. A notice board provides a visible place for shared announcements, while the sidebar leads to course management, student tracking, reports, and account settings.
The tracking screen filters students by course and presents student numbers, names, grades, and an action column. The UML extends this workflow with instructor methods for posting content, uploading grades, retrieving student tables, and generating course-specific reports.
For the administrative side, the design assigns Admin responsibility for creating student and instructor accounts, creating courses, assigning instructors, deleting records, and removing students from courses. This connects the CRUD requirement to explicit domain operations.
Instructor dashboard with course, student, and submission totals and a notice boardInstructor student tracker filtered by course, showing grades and student actions
06 / NOTIFICATIONS & REPORTS
Keep learners informed and make progress visible
The notification screen shows enrollment confirmations and grade-update messages, an unread count, and a Mark All as Read action. The design uses NotificationService and NotificationListener to connect changes in application state to student-facing updates.
The instructor report screen offers Student Performance Report and Course Activity Report options with a course selector. The UML also defines a ReportService for student, instructor, and course reports, and separates report generation from the screen that requests it.
Enrollment, grade updates, and activity submissions are the learning transactions represented in the model. Each connects a user action to changes in the relevant course, student, activity, or notification objects.
Student notifications for enrollments and grade updates with Mark All as ReadInstructor report selection with student performance and course activity options
07 / OOP & DESIGN PATTERNS
Give each object a clear responsibility
Singleton managers
CourseManager has a shared instance and a List of courses. AccountManager similarly centralizes user collections and the active account. These managers give the desktop forms a common source of application state within the running session.
Factory creation
ContentFactory defines creation methods for Content and Activity. Keeping construction behind a factory separates the form collecting input from the domain objects it creates. Video, Document, and Quiz were the content categories specified by the course brief; the supplied screens do not independently demonstrate every category.
Observer notifications
NotificationService keeps notification listeners and exposes notifyListeners(), while Student responds to notification updates. This gives the Observer requirement a concrete purpose: keeping the learning interface informed when relevant events occur.
Class relationships
Student-to-Course enrollment expresses association. Course collections of Content and Activity support the course-content aggregation model. Student and its profile illustrate composition, while the User hierarchy captures shared account identity with specialized behavior.
Loading diagram…
08 / MY CONTRIBUTION
Leading delivery and keeping quality visible
Overall leadership and quality assurance
I led the overall group project and handled quality assurance, coordinating delivery and checking that the LMS experience addressed the assignment requirements. The work we built connects role-based entry, a course stream, account management, an instructor dashboard, student tracking, notifications, and report selection. My focus was the coherence and quality of that complete experience across the team’s work.
Working within the data constraint
Using in-memory object lists made relationships and state changes central to the exercise. An enrollment, notification, grade, or content update had to be represented consistently across the relevant objects and screens without relying on a database to coordinate them.
Scope of the demonstration
NovaLearn was a classroom desktop prototype using demonstration data held in memory. Its value as a second-year project was the practical integration of GUI programming, OOP, and reusable design patterns, with leadership and quality assurance connecting the team’s individual work into a coherent deliverable.
09 / SCREENS
A look at the project
NovaLearn Student, Instructor, and Admin login selectionStudent course stream with dated learning contentStudent profile management and password actionsInstructor dashboard with course and student summariesCourse-specific student tracking and gradesStudent enrollment and grade notificationsInstructor performance and course activity report selection
1 / 7
10 / DELIVERABLE
What was built
A group-built Java Swing LMS prototype that brings course access, student profiles, instructor tracking, notifications, and report selection into one desktop experience. As Overall Project Lead with responsibility for Quality Assurance, I coordinated the group effort and helped keep the application aligned with the course requirements. The project demonstrates the transition from classroom OOP concepts to a connected application.
HTML5
CSS3
JavaScript
React
Next.js
Astro
Tailwind CSS
Java
Python
PHP
TypeScript
Node.js
MySQL
Firebase
Supabase
JSON
Vercel
Render
Git
GitHub
Docker
Figma
Canva
Codex
Claude
