# One Week of Learning at IT Del
#### Refined Draft by [[Hsueh-Yung Benjamin Koo|Benjamin Koo]], October 28, 2024
IT Del, the campus where [[Luhut Binsar Pandjaitan|Pak Luhut]] launched the nationwide learning initiative for the **GASing program** spearheaded by Prof. [[Yohanes Surya]], is positioned to pioneer a proactive, technology-enabled model of learning. As a technical university, IT Del has the unique potential to set a global standard by transforming GASing into a widely adopted practice, contributing a remarkable intellectual legacy from Indonesia.
After a week of engaging with students and instructors at [[IT Del]], I summarized the activities this week, focusing on content aligned with ongoing actions to extend this [[GASing]] initiative. The objective is to ensure that all students and teachers actively participate in technology-enabled learning environments. With the necessary tools for interactive and **AI-enhanced learning** already in place, the next immediate step is to bring more students and teachers on board.
To build an effective learning institution, universities must embrace [[GASing|GASing's]] motto: _Learning to Teach, and Teaching to Learn_. This vision is realized through two mutually reinforcing cycles: **Challenge Design** for Content Development and **Mission Execution** for Learning Operations. Each cycle serves as part of an iterative process, where development and operations continuously support and inform each other. To illustrate this, I presented the following diagram showing the [[DevOps]] workflow of [[XLP]], which represents this dynamic and adaptive approach to learning.
![[XLP_Workflow.png]]
See [XLP Workflow](https://docs.google.com/presentation/d/1-7g5UEt3iVdQI5k6YnjqddrUDcZj18RgRK2W6CwAPVo/edit#slide=id.g30d493dadd3_0_734)
### Universities as Institutions that can Learn
This workflow diagram outlines a universally applicable model that is especially relevant in the age of AI and Machine Learning. Designed as a reference framework, it supports the digitization of modern campuses and broader communities of learning. By leveraging the dual cycles of **Challenge Design** for Content Development and **Mission Execution** for Learning Operations, this model facilitates adaptive, data-driven approaches to teaching and learning.
In this structure, **Challenge Design** ensures that content development aligns with desired **Learning Outcomes**, while **Mission Execution** serves as a real-time, on-site feedback loop. The operational cycle provides immediate insights and **Learning Resources**, enabling content developers (mostly students) and instructors to adjust strategies based on real-world results. This adaptive, reflective approach is ideal for institutions seeking to implement AI and Machine Learning tools, offering a pathway for responsive, continuous improvement in educational environments.
Overall, this workflow provides a scalable, digital-first model for transforming campuses and learning communities into dynamically evolving, data-informed ecosystems.
# Teaching and Meeting Appointments
With Pak Luhut recently appointed as the **Advisor to the President on Digitization and Governance Technologies**, there is a growing emphasis on rapidly advancing content areas and the critical need for higher literacy in data sovereignty and network security protection. At IT Del, we are uniquely positioned to respond to these priorities by preparing students with the skills and knowledge needed to lead in these fields.
To support this mission, I was invited to present courses that align closely with these emerging areas:
- **Machine Learning**
- **Theory of Automata and Formal Languages**
- **Data Mining**
- **Student Projects** focused on [[Blockchain]]/[[Web3]] and Distributed Apps ([[dApps]]) development
Through these classes, we examined practical applications of these technologies and discussed how IT Del could leverage its unique position to develop expertise in data governance, security, and the sovereignty of digital assets. By embedding these principles into IT Del’s curriculum, the university can advance toward becoming a global leader, keeping pace with top technological institutions while meeting Indonesia's strategic goals for digital resilience and innovation.
## Original Schedule
A proposed schedule for the teaching assignments is outlined in the [[One Week of Learning at IT Del#Planned Schedule|Planned Schedule]] hyperlink.
### Lecturing as an art of story telling
Our team set a shared objective during Monday’s preparatory meeting: to create a dynamic, cohesive narrative for the week, immersing students and instructors in a learning experience that reveals the interconnectedness of intellectual concepts across courses. Through a series of carefully planned activities at IT Del, this approach fosters a multidisciplinary environment, allowing students to see how each subject enhances and complements the others.
The lecture slide sets, which form the basis for this **storytelling** approach, are publicly accessible on Google Drive:
- **Finite Automaton Theory**: [1U1nANZ2QsTQJxcDOtZqUnrmJrNCrp1-6](https://docs.google.com/presentation/d/e/2PACX-1vTgOxt8vFGkFuYvgFpik5VX9i9qZAXg65uNcA7ZU4N-dsvp9eLYg3hsUZDaO5m4MA/pub?start=false&loop=false&delayms=3000)
- **Machine Learning/Data Mining**: [1-7g5UEt3iVdQI5k6YnjqddrUDcZj18RgRK2W6CwAPVo](https://docs.google.com/presentation/d/1-7g5UEt3iVdQI5k6YnjqddrUDcZj18RgRK2W6CwAPVo/edit#slide=id.g30d493dadd3_0_77)
Serving as both instructor and active researcher, **[[Hsueh-Yung Benjamin Koo|Ben Koo]]** will lead this journey by integrating field research, programming, and teaching. This structured narrative ties together **[[One Week of Learning at IT Del#Theory of Automata and Formal Languages|Theory of Automata]]**, **[[One Week of Learning at IT Del#Machine Learning|Machine Learning]]**, and **[[One Week of Learning at IT Del#Data Mining|Data Mining]]**, highlighting the unity and progression of these topics. The framework, supported by interactive exercises and two slide sets, invites students to follow an **evolving storyline** in which a **common data pipeline** adapts across diverse applications.
Through this week-long immersion, students witness the organic growth of knowledge, as fields intersect, influence, and drive each other forward. Engaging with this unified intellectual thread transforms learning into a richly interconnected tapestry of ideas, crafting a meaningful and lasting educational experience.
## An Evolving Story Line from Automata Theory, Machine Learning, to Data Mining
The execution of the **story telling** plan at IT Del followed the original curriculum outline, ensuring that each activity reinforced the goals and topics within existing courses while introducing innovative, real-world applications to the students and faculty.
### Theory of Automata and Formal Languages
We began by establishing the foundational role of [[Formal Language]] and [[State Machine]] as core theories for data governance and complexity management at scale. Introducing the concept through the story of the [[Tower of Babel]], we highlighted the necessity of a shared intellectual foundation in formal languages to achieve engineering and organizational coherence. This led to discussions on the strategic power of state machines, which serve as meta-linguistic tools capable of modeling and analyzing complex systems across diverse applications. I first presented the story using the “Canvas” feature of [[Obsidian]], forming a "[[Finite State Diagram]]", which is the designated topic assigned to me:
#### Leveraging Finite State Machine for Story Telling
To illustrate that storytelling itself can be modeled as a Finite State Machine, I used Obsidian's **Canvas**, a **free-to-use** note-taking graphical tool. This allowed me to visually organize all the intricate details of the content through drag-and-drop functionality, with relevant media readily accessible at my fingertips, supporting a cohesive and interactive storytelling structure.
![[FiniteStateMachine_In_Canvas.png]]
#### Story Telling at the Transdisciplinary Level
To illustrate real-world applications, we delved into the origins of **[[Bitcoin]]** and **[[Blockchain]]**, both rooted in addressing the **[[Byzantine Generals Problem]]**—a challenge successfully tackled by Turing Award laureate **[[Leslie Lamport]]**. Lamport’s pioneering work on verifying the logical accuracy of **Finite State Machines** and developing domain-specific languages underscores the critical role of **Automata Theory** in today’s governance and digital trust frameworks.
The narrative continued with insights from my own research, where I developed an algebraic formalism (see **[[Algebra of Systems]]**) to define languages specifically for an engineering initiative. By applying **[[State Machine]]** principles, I engineered robust solutions for complex engineering problems. This journey included my visits to Dr. **[[Bob Seamans|Seamans']]** home—a key figure who famously assured President Kennedy that landing on the moon within a decade was achievable, as highlighted in the slide. During these visits, I gathered valuable oral history and discussed how Finite State Machines are instrumental in crafting effective, dependable engineering solutions. My thesis further demonstrated that Finite State Machines can be leveraged to solve intricate engineering challenges, particularly in Space Exploration programs.
![[TransdisciplinaryDecisions.png]]
These examples emphasized that effective management of formal languages is integral to building scalable, reliable systems—a critical capability in both technical and governance contexts.
Building on this foundation, we transitioned to modern applications in AI by discussing [[Large Language Models]], which can now generate complex state machine diagrams using tools like [[Mermaid]] and [[XState]]. These advancements make the principles of Formal Languages and State Machines directly relevant within the context of machine learning, illustrating their relevance within the domains of AI and data-driven governance. This clearly has direct implications in teaching and learning Computer Science/Inforamatics in general. For more details, please see [[@CrueltyReallyTeaching2022|On the Cruelty of Really Teaching Computer Science]] by [[Edsger Wybe Dijkstra]].
### Machine Learning
This session introduced students to cutting-edge machine learning tools, showcasing the power of locally fine-tuned large language models that can operate directly on laptops, even offline. I demonstrated **[[Fragments]]** and **[[Bolt.new]]**, which ran efficiently on my laptop and helped automate tasks such as Python coding for data processing. This included breaking down large images into smaller segments for effective image recognition, showing students how these tools could streamline developmental work without requiring constant internet access using locally run large language models through [[Ollama]].
These hands-on, offline machine learning tools provided timely and practical solutions for students’ projects and emphasized the potential to address local economic needs. This became particularly relevant through the **Crop Yield Analysis** project at **[[Lake Toba]]**. Using aerial images collected by **[[IT Del]]** graduates, we explored how localized machine learning solutions directly support regional industries, aligning perfectly with curriculum goals and giving students firsthand experience with impactful, real-world applications.
![[CropYieldAssessment_LakeToba.png]]
Despite technical challenges in connecting my laptop to the large screen, we managed to showcase several state-of-the-art machine learning applications in the classroom. This logistical issue also emphasized the need to avoid information isolation by streamlining device connectivity. Connecting smart screens through an [[Overlay Virtual Private Network]] ([[Overlay VPN]]) could provide a systematic solution, reducing the need for last-minute wiring and enabling data sets and content to be preloaded and tested remotely. This experience inspired me to consider deploying Overlay VPN technology on campus to enhance connectivity, improve content accessibility, and ensure that technical setups do not hinder learning. This approach would allow IT Del to remain at the forefront of seamless, integrated learning environments.
### Data Mining
Throughout three sessions with the **Data Mining** class this week, I guided students to understand the core principles of data governance, captured in the slogan: **[[Timeliness|Timely Feedback]]**, **[[Accountability|Accountable data sources]]**, and **[[Observability|Observable outcomes]]**. The main takeaway was that the true value of data lies in these three qualities, which ensure that data mining efforts are not only effective but also actionable and reliable.
#### Identifying Hidden Opportunities using LDA and Overaly VPN
Using a personal dataset of over 10,000 notes, I demonstrated how **[[Latent Dirichlet Allocation]]** ([[LDA]]) can automatically categorize these notes by associating them with a specified number of topics. This showcases how content creators can leverage LDA to extract meaningful insights from large collections of **self-administered content**. By applying LDA, students at IT Del were able to categorize their study notes into key themes using just a laptop and a stable network connection, making data analysis both accessible and directly relevant to personal learning. This hands-on experience illustrated the practical applications of data mining, showing students the value of organizing and understanding their own data in a meaningful way.
However, the process highlighted infrastructure limitations: downloading the 3 Gb **Natural Language Toolkit** (NLTK) dataset took over ten hours, emphasizing the need for a robust **[[Content Distribution Network]]** ([[CDN]]) on campus. A reliable CDN would significantly improve access to large datasets, supporting the adoption of advanced data mining tools like LDA. Such infrastructure would empower students and researchers to uncover hidden value in personal data assets efficiently, unlocking deeper insights and accelerating learning across the campus.
To address these needs, we discussed strategies for network optimization. The teaching assistant, already exploring **IPFS** technology as part of the **Blockchain/Smart Contract Student Project**, demonstrated the potential for decentralized content sharing. As a hands-on exercise, we also set up Overlay VPNs using [Tailscale.COM](http://tailscale.com)(other solutions includes [Netbird.IO](http://netbird.io) and [Zerotier.ORG](http://zerotier.org) ), enabling all 41 students to create personal private networks across multiple devices in under two hours. This integration of real-world technology and curriculum content not only underscored the importance of **Timeliness**, **Accountability**, and **Observability** but also equipped students with future-focused skills for effective data governance.
###### Students showing off their Network Connection Successes
![[OVPN_at_ITDEL_Observable.jpeg]]
# Student Projects
Working alongside Prof. Johannes Sianipar, I presented my work in developing the **[[Extreme Learning Process]]** ([[XLP]])—a method adopted by top institutions such as [Tsinghua](https://2018.govtechbrasil.org.br/speaker/benjamin-koo/?lang=en) and [MIT](http://roadmaps.mit.edu) for its learning-centered approach. Known for its international acclaim, [[XLP]] integrates information technologies to verify and document student-driven projects, fostering both accountability and transparency in learning activities.
## The Geographic Distribution of Data Sources in Indonesia
Indonesia’s unique **geographic distribution** offers a powerful opportunity to advance its digital infrastructure by applying **Blockchain** and **Smart Contract** technologies, tailored to reflect its dispersed landscape. Building on foundational concepts covered in class, IT Del is positioned to explore an often-overlooked aspect of Blockchain: the integration of **spatial data structures**. These structures, which encompass physical data storage and geographical distribution, can lead Indonesia in developing Blockchain applications that address the needs of a vast archipelago, enhancing governance and fostering data-driven transparency on a national scale.
Combining **[[IPFS]]**, Blockchain, and Smart Contracts with advances in Software-Defined Networking, particularly **Overlay Virtual Private Networking** (as demonstrated in our Data Mining class), IT Del can create a versatile spatial-temporal data processing pipeline. This setup would support secure, distributed data sharing across various locations, promoting trust and interoperability between communities locally and globally.
This pipeline’s potential aligns perfectly with the broader goals of **GASing** as articulated by Prof. Surya, which can expand beyond IT Del, providing an adaptable framework that could scale to any learning institution worldwide. By leveraging the prior success of GASing, IT Del may become the model institution that leverages **open-source, self-adaptive learning instruments** at an internet scale, transforming educational settings into dynamic, community-centered networks with solid theoretical underpinnings and robust technological foundations.
Through the [[ABC curriculum]] and [[XLP]] workflow, students are encouraged to move beyond traditional lectures, engaging in self-directed, cross-disciplinary projects. Empowered by Overlay VPNs and decentralized networks, students experiment, collaborate, and build applications that meet local community needs. This integrated framework not only supports IT Del’s curriculum but also offers a blueprint adaptable to other educational institutions. This approach envisions Indonesia as a leader in technological education, equipping students with critical AI and Machine Learning skills that are globally relevant and empowering the nation to leapfrog into the age of advanced digital technologies.
With a diverse student body of 1,600 spanning multiple disciplines, IT Del has a significant opportunity to drive impactful research and development. By utilizing Overlay VPNs tested in the Data Mining class, students can implement spatially distributed Blockchain solutions, addressing governance and data security needs across Indonesia. This innovative approach positions IT Del at the forefront of developing spatially aware Blockchain applications, making a meaningful impact that leverages Indonesia’s unique geographic distribution while contributing to the global knowledge economy.
# Extracurricular Activities
Inspired by Prof. Sianipar's vision for transformative interdisciplinary education, I partnered with him to launch the **Extreme Learning Process** (**XLP**) at IT Del. To foster engagement, I reached out to **Stefhani Kezia Siburian**, a motivated Biology student who encouraged students from various fields to participate in this educational network. She also collaborated with **Ivana Godelva**, a well-connected peer, to build an interconnected network of student organizations using **Overlay VPN** technology. This created a dynamic, collaborative environment where students from diverse disciplines joined forces, with eight students promptly setting up personal VPNs. This setup enabled smooth, secure communication across devices connected via local and global networks.
**Overlay VPN technology** is a critical tool that enables **[[software-defined networking]]** infrastructure. It connects devices effortlessly across local or multiple networks, overcoming typical barriers to installation and learning. This accessibility makes Overlay VPN an ideal foundation for IT Del’s campus, positioning it to **leapfrog** other educational institutions by integrating network security and connectivity advancements into the curriculum. With formal instruction in these technologies at an institutional level, IT Del could empower students with cutting-edge skills, fostering a campus environment that is secure, interconnected, and future-ready. This initiative has also seen students connecting with peers in the Data Mining class, sharing knowledge and drawing on each other’s strengths to further solidify IT Del's role as a leader in digital education.
###### A Picture taken after the Data Mining Class
![[DataMiningClass_ITDel.png]]
This cross-disciplinary initiative encourages students to realize the importance of social and digital connectivity, redefining their learning experience at IT Del. By using locally operated Large Language Models and forming partnerships across departments, students will uncover previously unexpected ways to connect, problem-solve, and broaden their academic horizons—epitomizing the essential role of cross-pollination on campus. This spirit of collaboration aligns with the goals of the **GASing** movement, which began with Prof. Surya and Pak Luhut and was revitalized at IT Del, empowering students to harness the full potential of interdisciplinary learning for a transformative educational journey.
### A Discussion about Del Bot
To demonstrate the feasibility of using locally trained large language models, I presented a prototype chatbot, temporarily named **Del Bot**. This demo showcased how students could be used to train specialized models—such as the Yolo 8 model I trained to recognize cabbages (see the demo for Machine Learning class)—and how additional datasets from Hugging Face or locally collected data can enable training on personal computers for a fully functional chatbot.
![[DelBot_2024_10_21.png]]
This demo highlighted three key features relevant to the learning experience I aim to bring to the campus:
1. **Overlay Virtual Private Network (VPN)**: Using an [[Overlay VPN]] (specifically [[Zerotier]]) and a single static IP address from **Amazon Web Services (AWS)**, I configured a reverse proxy on a dedicated machine. This setup supports unlimited domain name declarations, allowing users to assign custom domains and publish computational services with ease. For example, I used the domain **astro.xlp.pub** to make the service publicly accessible across the Internet, enabling open access and interaction.
2. **Responsive UI with "Astro" Content-First Framework**: By adopting the [[Astro]] JavaScript framework, I developed a web-based app that supports a responsive UI, adapting seamlessly across devices, including tablets, desktops, and cell phones, with consistent, user-friendly behavior.
3. **Locally Trained Large Language Model**: The locally fine-tuned model preserves multilingual capabilities, including **Bahasa Indonesia**, enabling the chatbot to deliver subject-specific knowledge while also supporting content in locally relevant languages.
This demonstration emphasized how campus-based projects can leverage global computing power and open-sourced data resources to create adaptable, user-centered applications tailored for educational and community use.
### Building a Campus That Connects Beyond Boundaries
On Thursday, Professor Sianipar gathered many instructors for a memorable dinner at Bistro by the Lake. This gathering was truly enjoyable—not only for the excellent food but also for the chance to engage with colleagues in a relaxed setting. I’m grateful for this thoughtful arrangement. The following day, I invited students to join me for an off-campus dinner, encouraging them to break from routine and engage with the broader community. Yet, transportation challenges often limit such outings, underscoring the need for secure, reliable travel options. Collaborations with local providers or community-backed services could offer safe, convenient transport for students, faculty, and visitors. As IT Del’s reputation grows, dependable transportation will be crucial to hosting distinguished guests and enhancing the campus’s appeal within the picturesque Lake Toba area.
![[DinnerOutings.png]]
# Conclusion: The Mantra of the Science of Governance
These activities not only enhance IT Del’s curriculum but also expand it, integrating practical applications and foundational skills from cutting-edge fields to enrich students' educational journeys. Each session reinforces IT Del’s mission to embed advanced technologies within its programs, preparing students for success in a digital world and empowering them to lead. This holistic approach allows us to introduce non-IT students to essential digital literacy skills, fostering an interdisciplinary environment that transcends traditional boundaries. Key competencies like **[[Timeliness]]**, **[[Accountability]]**, and **[[Observability]]**—the core of The Science of Governance—are central to this mission.
![[Time_Account_Observe.png]]
In alignment with IT Del's focus on Informatics, this curriculum integrates Formal Language and Automaton Theory, Machine Learning, Data Mining, and open-source data and software packages, including Overlay VPN technologies. Together, these components illustrate how a modern, adaptable curriculum can be applied and continuously improved at IT Del. Leveraging the experience of deploying GASing as a nationwide program, we are positioned to further refine this initiative within the institution. By utilizing advanced networking technologies, IT Del can ensure that the goals of delivering impactful, data-driven governance solutions that extend beyond the campus, shaping practices across the nation and marking IT Del as a hub for transformative digital progress.
Under Pak Luhut’s vision for institutions advancing digitization and governance, IT Del stands at the forefront of this technology-enabled **Science of Governance**—a journey that started here at Lake Toba and promises exciting contributions on a national and global scale.
---
# References
```dataview
Table title as Title, authors as Authors
where contains(subject, "Lecture Outline") or contains(subject, "GASing") or contains(subject, "Teaching Computer Science")
sort title, authors, modified
```
---
# Appendix
### Planned Schedule
[The original Google Sheet](https://docs.google.com/spreadsheets/d/1-J8pmzLXmPoE4T0t96Wr59a8VqK8KKoJ/edit?gid=1419955803#gid=1419955803)
![[October_ITDel_Schedule.png]]