Loading...
Last updated: September 9, 2026
Kernpool is operated by Light & Honor Technologies Pte. Ltd. (UEN 202245535R), registered in Singapore.
For the information described in this policy — your account, your billing records, the metadata about what we scheduled — that company is the data controller. The company is Singaporean; the servers are not necessarily. Section 7 says where the data sits.
There is a second case worth separating out. Your job names and the logs you choose to stream to your dashboard may contain personal data about your own users or staff, which you put there. For that data you are the controller and we act on your instructions, as a processor. We only hold it to show it back to you, we delete it on the schedule in section 8, and we do not use it for anything of our own. If you need processor terms for your own compliance, write to us and we will provide them.
Write to [email protected] with any question about this policy or to exercise a right in section 9. That address also reaches the person here responsible for data protection.
You give us
The Service records
We do not collect
| Purpose | Data | Basis |
|---|---|---|
| Run your account | Account record | Contract |
| Run and display your jobs | Job metadata and logs | Contract — and where these contain personal data about your own people, on your instructions as processor (section 1) |
| Meter and bill | What we scheduled, and payment records | Contract |
| Answer your support messages | Support correspondence | Contract / legitimate interest |
| Detect abuse and secure the network | IP, sign-in events, job patterns | Legitimate interest |
| Check a billing dispute against what we actually launched | Scheduling logs | Contract, and our legitimate interest in being able to check a claim |
| Meet tax and accounting duties | Invoices, fee records | Legal obligation |
| Check a refund is going to the right person, and meet financial-crime duties that apply to us | Registration and sign-in record, payment records | Legitimate interest, and legal obligation where one applies |
| Product and service email | Email address | Contract / consent |
We do not use your data to train our own models, and we do not sell it.
Your code, datasets, and model weights never reach our servers.
Kernpool schedules work onto machines inside your cloud accounts and clusters. Your workload is submitted to that machine and runs there. We see the instruction to launch it, and the metadata that comes back — job name, status, duration, resource footprint, and any logs you choose to stream to your dashboard.
We do not copy your training data. We do not read your model weights. We do not use anything inside your workloads to train our own models. The machine belongs to your account, under your provider's terms and your own IAM policy — there is no host of ours in between.
What this also means: the security of that environment is yours to set. Encryption at rest, network rules, and who in your organisation can reach the machine are configured in your cloud account, not in Kernpool.
If the business is sold. Account data may transfer to a buyer. We will tell you at least 30 days before that happens. A change of owner is not something you chose, so you do not have to give up a balance to avoid it: section 6 of the payment and refund policy sets out the refund and the deadline for asking.
To schedule into your accounts, Kernpool holds whatever access you grant — an IAM role, a service account key, a kubeconfig. This is the most sensitive thing we store, so it is worth being specific.
Your account, your billing records, and the metadata about what we scheduled are processed and stored on servers in the United States, and may be processed in other countries where our infrastructure providers operate. If that changes materially we will update this page.
Your workloads and their data are a separate matter entirely. They stay in whichever region of your own cloud account they were scheduled into, which you control and we do not touch. Nothing in this section is about them.
If you are in the EEA, the UK, or Switzerland, the account data described in section 2 leaves your region when you sign up, wherever it lands. Transfers rely on appropriate safeguards, and only under a legally valid transfer mechanism.
| Data | Retention |
|---|---|
| Account record | Until the account closes, then 30 days |
| Job metadata and logs | 90 days. Logs you chose to stream to the dashboard are on the same clock — we are not a log archive, so keep your own copy if you need one for longer. |
| Cloud credentials | Until the earliest of: you revoke them, 24 months of inactivity, or 7 days after the account closes. |
| Scheduling logs — what we launched, when, and on whose instruction | 24 months. These are what we check a billing dispute against, so they outlast the job logs above. |
| Support correspondence | 24 months after the conversation ends |
| Workload contents | Never held by us — governed by your own cloud retention settings |
| Billing and tax records | As required by law, typically 5–7 years |
| Security logs | 180 days |
Only the account record and your cloud credentials are tied to closure. Everything else on a rolling clock — job logs, scheduling logs, support correspondence, security logs — runs to its own schedule whether the account is open or not.
Accounts that go quiet. We do not close an account for inactivity, but we do not keep everything either. After 24 months with no activity we delete the connected-account list and the cloud credentials themselves — neither is doing anything for you at that point, and your job metadata and logs will already have gone on their own 90-day clock. The scheduling logs run their own 24-month clock and go at the same time. What stays is the minimum needed to keep the account yours and to pay out a refund in the cases where one becomes due: your email address, the balance, the payment records we are required to hold, and enough of the original registration and sign-in record to confirm it is really you and to satisfy whatever financial-crime duties apply to us. We keep that last part because a refund years later still has to reach the right person; the basis for it is our legitimate interest in not paying out to the wrong claimant, together with any legal obligation that applies, and not your continued use of the Service. Come back and the account still works — you will just be reconnecting your clouds from scratch.
You can ask us to:
One thing to flag: some of what we hold is personal data about your people, inside job names and the logs you chose to stream (section 1). We hold that on your instructions, so if one of your own users asks us for a copy or a deletion, we will point them at you rather than act on it ourselves, and tell you it happened.
Write to [email protected]. We respond within 30 days. If you are not satisfied you can complain to the PDPC.
No system is perfectly secure. If a breach affects your data we will tell you as soon as practicable, and we will notify the PDPC within 3 calendar days of concluding that it is a notifiable breach, as the PDPA requires.
We use a session cookie to keep you signed in, and a small number of cookies that remember preferences. There are no advertising or cross-site tracking cookies. Blocking the session cookie will stop you being able to sign in.
The Service is not for anyone under 18. We do not knowingly collect data from children. If you believe a child has created an account, tell us and we will delete it.
We will post changes here and update the date at the top. For material changes we will email you at least 30 days before they take effect.
Light & Honor Technologies Pte. Ltd.
UEN 202245535R
[email protected]