SK hynix IT roles evolve into digital brains for AI manufacturing
SK hynix is evolving its IT roles from simple system operations to "digital brains" that lead AI transformation (AX) by optimizing semiconductor manufacturing through AI-driven data pipelines and internal Large Language Models (LLM).
IT roles at SK hynix are transforming into "digital brains" to lead AI transformation by managing data pipelines and optimizing semiconductor production. Using RAG technology, an internal LLM service now allows users to simultaneously reference approximately 200 documents per laptop to improve work efficiency.
The key to determining the efficiency of semiconductor manufacturing processes lies in how massive amounts of data are managed and utilized. With the arrival of the AI era, IT roles at semiconductor companies are evolving beyond simple system operations into "digital brains" that lead enterprise-wide AI transformation (AX).
A video released by SK hynix shows the practical work sites of IT roles, which handle everything from planning 24-hour IT infrastructure to lead the AI era, to building manufacturing data pipelines and developing internal Large Language Model (LLM) services. The video demonstrates that IT roles are not limited to simple coding but perform a core role in optimizing the physical flow of semiconductor production using AI.
From AMHS Optimization to Failure Prediction
Inside semiconductor plants, the Automated Material Handling System (AMHS), which transports wafers, moves incessantly. The core personnel of the IT role are those who develop AI algorithms to ensure this system moves along optimal paths without congestion and operate systems that predict failures.
Lee Ji-ho, a TL, stated that she is in charge of AMHS intelligence work and manages systems directly linked to fabs at campuses where actual plants are located, such as Icheon and Cheongju. "Unlike tech companies, IT roles at semiconductor companies involve many tasks close to the field," she explained. "The role is to create the foundation upon which AI can function well by supporting the data and technology required for R&D, the field, and operations." Currently, SK hynix has actual plants at its Icheon and Cheongju campuses, a system-centered workplace in Bundang, and a new campus under construction in Yongin. At Icheon and Cheongju, staff manage systems directly linked to production next to the fabs, while systems not directly connected to production are mainly managed in Bundang.
In the actual process field, logistics equipment called Overhead Hoist Transport (OHT) moves. While humans previously intervened or equipment moved according to simple rules, AI algorithms now minimize congestion zones and find optimal paths. According to a simulation in the video, the goal is for the entire system to move efficiently without congestion, rather than having a single specific piece of equipment run at maximum speed. This principle is similar to how dispersing overall traffic flow is more important than the speed of individual vehicles during holiday road congestion. Additionally, using AI to analyze causes and derive countermeasures when failures occur is one of the major tasks of the IT role. Currently, the infrastructure integrated control room at the Icheon campus monitors and controls infrastructure control on the 2nd floor and production-linked systems on the 6th floor in real time.
Internal LLM Secures Security and Efficiency via 'RAG' Technology
Recently, there has been a strong movement within SK hynix to improve productivity using AI. In particular, to suit the characteristics of the semiconductor industry where security is vital, an internal LLM service has been established to allow the safe use of internal information without leaking data to the outside. It is designed to maximize the work efficiency of members through AI while preventing the leakage of confidential information and document security breaches.
The internal LLM provides various functions such as document summarization, report writing support, mind map generation, and text-to-image generation. The core of this service is 'RAG (Retrieval-Augmented Generation)' technology. When a user asks a question, the AI finds relevant knowledge from the internal database to generate an answer, thereby increasing accuracy. The video shows the process of how this technology improves the quality of answers by retrieving existing knowledge from the database when a user asks a question.
Baek Won-gi, a TL, said, "It can refer to not just one document, but approximately 200 documents per laptop to provide an answer simultaneously." He added, "In the past, this was an area exclusively handled by IT roles, but now the situation is changing to the point where, if the foundation system is provided, business departments (such as industrial roles or device development) can create and use the agents they need themselves." In fact, members are directly creating and utilizing AI agents specialized for their own work based on the foundation provided by the IT department.
Requirements for IT Talent in the AI Era: "Fundamentals and Field Understanding"
What competencies are required for IT talent in an era where AI performs coding? Representatives of the IT roles at SK hynix cited 'coding fundamentals' and 'understanding of semiconductor flow.' This is because solid basic capabilities are essential to review and understand code generated by AI. A representative advised, "You must know the fundamentals for coding to understand the code created by AI. It is necessary to gain experience working efficiently through AI projects and to have the capability to consider how to increase work efficiency using AI."
Furthermore, as IT systems are closely connected to the production field, the ability to accurately grasp the requirements of the field is important. This is because if the flow of the field is not understood, mistakes in the system can lead directly to product issues or decreased production efficiency. The representative emphasized, "It is important to take an interest in the overall flow of semiconductors, and since they handle systems related to production, they must continuously listen to the field to understand the flow."
The infrastructure integrated control room at the Icheon campus is expected to expand further alongside the construction of the new Yongin campus. It is expected that an even larger-scale control facility, integrating AMHS and system control, will be built at the Yongin campus. IT infrastructure armed with AI and data is supporting the foundation of semiconductor manufacturing and accelerating SK hynix's AX (AI transformation).
Source: original video (YouTube)
0Comments
Comments are currently disabled.