Among Korean cloud service providers (CSPs), NAVER Cloud is the name developers mention most. For anyone who says “I want to be a cloud engineer” in Korea, it is one of the top targets.
So I opened every open posting on its careers site to see what it is actually hiring for. That is 7 postings, or 11 roles once you count the sub-roles inside them.
The short version: in NAVER Cloud’s current postings, the word “cloud” barely appears. What appears is security and AI. This post covers why, and what it means for people aiming at cloud careers.
⏰ Closing soon: all 7 postings close on October 14, 2026. Experienced roles close at 5 p.m. and internships at 6 p.m. (KST). If a role interests you, check the posting before reading the rest of this.
At a glance: 11 roles
| Type | Role | Field | Experience required |
|---|---|---|---|
| Experienced | Security Solution Operations (Junior Talent) | Security | 2–4 years |
| Experienced | AI Security Training Data Engineer | Security × AI | 3+ years |
| Experienced | AI Security RL Environment Engineer | Security × AI | 3+ years |
| Experienced | Offensive/Defensive AI Security Engineer | Security × AI | 3+ years |
| Experienced | Security for AI Engineer | Security × AI | 3+ years |
| Experienced | AI Security FDE Engineer | Security × AI | 3+ years |
| Intern | Cloud IaaS Operations Planning | Cloud | Bachelor’s |
| Intern | AI Model and Agent Systems Research | AI | Master’s/PhD |
| Intern | World Model and Image Generation Model Research | AI | Master’s/PhD |
| Intern | World Model and Image Generation Model ML Engineer | AI | Bachelor’s/Master’s |
| Intern | Generative AI Content Creation and Automation | AI services | Bachelor’s |
In numbers:
- 6 experienced roles = 6 security roles. Five of them were posted together by a single “AI security” team.
- 5 internships = 4 AI + 1 cloud. All internships are experience-type (not a path to a full-time offer) and run for about three months starting November 2026.
- Every role is based at Green Factory in Bundang, Seongnam.
Now one by one, starting with what is closest to cloud careers.
① The only posting with “cloud” in its name: IaaS operations planning internship
IaaS (Infrastructure as a Service) is a cloud product that rents out basic infrastructure such as servers, storage and networking. AWS EC2 and NAVER Cloud’s server products fall in this category.
This department handles infrastructure management for all NAVER Cloud products (SIM, Service Infra Manager) and IaaS product planning. It describes itself as leading new product development, infrastructure architecture, structural improvement and stabilization.
What the intern actually does, though, is quite specific.
- Plan and build a new IDC floor space usage dashboard
- Gather requirements by interviewing team members and write screen layouts and feature definitions
- Develop features, refine the screens and write a user guide
Floor space here means the floor area in a data center where server racks sit. It is managed in units like one rack position or so many kW of power. In other words, this intern builds “an internal tool that shows at a glance how much data center space is used, and how.”
The requirements stand out. Experience developing and running a web service with React and FastAPI is mandatory. Docker and Python data processing are preferred.
The infrastructure planning team at a cloud company is hiring a web developer, and that is the key point of this posting. A large part of cloud operations is building your own “tools that make the current state visible.” The people working behind the console end up writing dashboards too.
② The core of the experienced roles: five jobs on the “AI security” team
This is the biggest block in the current postings. One posting bundles five sub-roles.
The team describes itself along two axes: AI for Security (using AI to strengthen security) and Security for AI (keeping AI itself safe). Its biggest current task is this:
As the lead institution for the “Development of Specialized AI Foundation Models (Cybersecurity)” program of the Ministry of Science and ICT and the National IT Industry Promotion Agency, the team aims to build an AI model specialized for the security domain in-house and release the results as open source.
A foundation model is a large base model that can be adapted to many uses. So this team is leading a government-funded project to build a security-specific AI model, and it is hiring the people needed to build it all at once.
What is interesting is how the roles are divided. The model-building process is split into stages, with one role assigned to each.
| Role | Stage of model development | One-line summary |
|---|---|---|
| Training Data Engineer | What to teach | Decides whether security data can be used for training and sets the labeling standard |
| RL Environment Engineer | What to test with | Builds isolated environments where the AI attempts security tasks, plus automatic graders |
| Offensive/Defensive AI Security Engineer | What to have it do | Designs AI agents that carry out attack and defense |
| Security for AI Engineer | Is it safe to release | Attacks the model and decides the scope and conditions of release |
| FDE Engineer | How to deploy it in the field | Installs and validates it on site at air-gapped institutions |
The posting states that the model training itself is done in collaboration with the AI model development organization. This team is not the one pressing the train button; it designs the data, the test environments and the evaluation criteria.
The role cloud-minded candidates should look at: RL Environment Engineer
Of the five, this is the one that touches cloud and infrastructure skills most directly.
RL (Reinforcement Learning) is a way for AI to learn by attempting tasks and receiving a score (reward) based on the outcome. For security tasks, that means asking it to “reproduce this vulnerability” or “write a rule that detects this attack,” then grading success automatically.
That requires thousands of isolated execution environments where the AI is free to attack. This role builds them.
- Design and operate sandboxes (isolated execution environments)
- IaC-based environment provisioning, packaging, version control and reproducibility
- Manage concurrent execution throughput across many sandboxes, and keep reset and restore stable
- Block reward hacking, where the AI fools the grader, and attempts to escape isolation
IaC (Infrastructure as Code) means writing server and network configuration as code and building it automatically instead of clicking through consoles. Terraform and Ansible are the best-known tools. Reward hacking is when an AI exploits gaps in the grading method to collect points instead of actually solving the task.
The mandatory requirements are experience building and operating isolated environments such as containers, VMs and sandboxes, plus Python. The preferred list puts large-scale Kubernetes operations and Terraform/Ansible side by side with the reinforcement learning serving stack (Ray, vLLM and so on).
Read closely, this is effectively a posting for “a cloud engineer who runs infrastructure for AI training.” The barrier is that three years of hands-on security experience are mandatory.
The other four roles in brief
The Training Data Engineer looks at security data such as malware, traffic, monitoring logs and vulnerability information and decides whether to include it in training and how to label it. The role also designs the data mix for each training stage (CPT, SFT, RLVR). Hands-on experience judging “this is a real attack / a false positive” in security monitoring or incident response is mandatory.
The Offensive/Defensive AI Security Engineer ties together models, security tools and execution environments to build AI agents that carry out attack and defense. The preferred list includes implementing MCP (a standard protocol for AI to call external tools) and defending against prompt injection.
The Security for AI Engineer is the person who attacks the AI instead. The role tests jailbreaks, training data leakage and misuse of agent permissions, and uses the results to decide how far the model can be released. It also defines release artifacts such as model cards and SBOMs. Frameworks such as NIST AI RMF, MITRE ATLAS and the OWASP Top 10 for LLM are named one after another.
The FDE in FDE Engineer stands for Forward Deployed Engineer, an engineer who goes on site and builds things at the customer. The role installs and validates security AI in the network-separated and air-gapped environments of critical-infrastructure institutions in power, telecom, finance, the public sector and aerospace. Preferred qualifications include building on-premises installation packages that include inference on Korean-made AI chips, and handling the national network security framework (N2SF). This is a role where a cloud company carries AI into places that are not cloud (air-gapped networks).
③ One more experienced security role: Security Solution Operations (Junior Talent)
This is a separate posting from the AI security team. At 2–4 years of experience, it is the experienced role with the lowest barrier to entry in the current postings.
The department handles technical security controls that protect the infrastructure and work environment of all of Team NAVER. The work includes:
- Protecting endpoints such as employee PCs, servers, containers and hypervisors
- Operating EPP/EDR (detecting and responding to malicious behavior on PCs and servers), PMS (patch management) and CCE (configuration vulnerability checks)
- Assessing user and device posture in a Zero Trust environment
- Security operations for AI agent and AI tool usage
- Automating security solution operations with AI engineering
Zero Trust is a security model that drops the assumption “if it’s inside the corporate network, trust it” and verifies every connection every time.
From a cloud perspective, the preferred qualifications are what to watch. They include experience migrating on-premises environments to the cloud or operating them, and experience configuring AI agent execution environments, integrating tools and building evaluation systems (harness engineering). Cloud migration and running AI agents are starting to attach to security operations roles as baseline skills.
④ Four internships: AI research and engineering
These are not cloud roles, but they show that NAVER Cloud is also an AI infrastructure company, so here is a short summary.
| Internship | What it does | Link to cloud and infrastructure |
|---|---|---|
| AI Model and Agent Systems Research | Multi-agent collaboration, decision models, multimodal model analysis | The team itself works on LLM inference technology (PEFT serving, speculative decoding, agent scheduling). Preferred: vLLM, SGLang, GPU performance analysis |
| World Model and Image Generation Model Research | Research on diffusion-based generative models and world models | Pretraining on large GPU clusters, distributed training infrastructure |
| World Model and Image Generation Model ML Engineer | Training data pipelines, synthetic data, evaluation automation | Mandatory: understanding of compute and I/O bottlenecks in large-scale data processing |
| Generative AI Content Creation and Automation | Writing AI task instructions (Skills), designing output evaluation criteria | AI agent services built on HyperCLOVA X |
Speculative decoding is an inference technique in which a small model guesses ahead and a large model verifies the guesses in one pass, producing answers faster. Having a team that researches this means that at NAVER Cloud, AI inference cost and speed are infrastructure problems.
The ML engineer internship listing “an understanding of compute and I/O bottlenecks in large-scale data processing” as mandatory fits the same picture. If training data can’t be fed to the GPUs in time, expensive GPUs sit idle. We covered this bottleneck in detail in GPUs by the thousands waiting on storage.
What these postings tell us
Three points.
First, NAVER Cloud’s hiring center of gravity right now is “security × AI.” All 6 experienced roles are in security, and 5 of them are for a government project to build a security-specialized AI model. This is likely a miniature of the company’s overall strategy. For a Korean CSP, the public sector, finance and critical infrastructure are core markets, and those markets cannot be entered without security certification and air-gapped deployments.
Second, the absence of a traditional “cloud engineer” posting doesn’t mean cloud skills aren’t needed. The RL environment posting asks for Kubernetes, Terraform and isolated-environment operations; the security operations posting asks for cloud migration experience; the IaaS internship asks for web development and Docker. Cloud skills are dissolving into baseline requirements for other roles rather than standing as a separate job. We saw the same trend in domestic MLOps job postings at Kakao, Coupang and others (in Korean).
Third, “running AI agents” has become a new common requirement. The security operations posting includes “security operations for AI agent usage” and “harness engineering”; the content internship includes “designing AI task instructions (Skills)”; and three of the AI security roles mention MCP and tool calling. Regardless of job family, companies have started asking about the ability to attach, control and evaluate AI agents.
If you are preparing to be a cloud engineer, the combination to take from these postings is this: the ability to stamp out isolated environments (containers, VMs) as code (IaC) and run AI agents safely on top of them. Gather the requirements scattered across NAVER Cloud’s current postings into one, and that is the shape you get.
⚠️ This post summarizes the 7 open postings and their 5 sub-roles on NAVER Cloud’s careers site (recruit.navercloudcorp.com), each opened and read directly as of October 12, 2026. Postings open and close at any time, so this snapshot does not represent the company’s hiring for the whole year. “No experienced-hire posting for a traditional cloud engineer” is an observation limited to the postings open at this moment. Role descriptions and requirements are summaries of the original postings; always check the original before applying. Postings may close early or be extended.
Frequently asked questions
When do the NAVER Cloud postings close?
All 7 postings open as of October 12, 2026 close on October 14. The experienced-hire postings (security solution operations and the AI security roles) close at 5 p.m., and the 5 experience-type internship postings close at 6 p.m. (KST). The postings say they may close early once candidates are selected, or be extended if needed.
Is there nowhere for an entry-level cloud engineer to apply right now?
There is no entry-level full-time posting at this point. Entry-level candidates can apply to the 5 experience-type internships, and the one directly related to cloud is the “Cloud IaaS Operations Planning” internship. It is open to current undergraduates or graduates, and experience developing and running a web service with React and FastAPI is mandatory. Because it is experience-type, the contract ends when the internship ends.
Do experience-type internships convert to full-time roles?
The posting states that it is “an experience-type internship, not a hiring-track one, and the contract ends when the internship period ends.” It does not assume conversion to a full-time role. That said, having worked with a Korean CSP’s real products and GPU environments is a strong credential for later applications.
Can I apply to more than one of the five AI security roles?
No. The posting says “duplicate applications are not allowed; you may apply to another position once the process for the position you applied to has fully concluded.” The five roles ask for quite different experience (data judgment, isolated-environment building, agent development, AI red teaming, on-site deployment), so you need to pick the one that best fits your background.
Can I apply to the RL Environment Engineer role without security experience?
According to the posting, it would be difficult. At least three years of hands-on experience in one of security infrastructure, vulnerability analysis, penetration testing or security tool development is mandatory, plus either experience building security practice environments such as CTF infrastructure or cyber ranges, or experience implementing logic that automatically judges execution results. Kubernetes and Terraform experience alone is not enough, but the role is a good fit for infrastructure engineers who do have a security background.
How does the experienced-hire process work?
For the experienced security postings, the steps are document screening (including a company culture fit assessment and a job test), a first interview, reference checks with a second interview, compensation negotiation and the final offer. There is a three-month probation period after joining. Only applicants who have completed or are exempt from military service may apply, and applicants must have no disqualifying reason for overseas business travel.
Leave a Reply