Xtecq provides RLHF, AI evaluation, and multilingual annotation in English, Bahasa Melayu, Mandarin, and Cantonese — with the cultural depth that generic vendors can't match.
What We Handle
RLHF & Preference Ranking
EN · BM · Mandarin · Cantonese
AI Output Evaluation
Accuracy · Tone · Cultural fit
Data Annotation & Collection
Text · Audio · Image · Video
🚀 Free Pilot Available
Start small. Verify quality. Scale when ready.
With 680+ million people across dozens of languages and dialects, SEA requires annotation teams who genuinely understand the region — not vendors who treat Malay as a simple translation exercise.
Religious sensitivities, local humor, and contextual idioms are invisible to generic annotation tools and offshore teams.
Malaysians naturally mix Malay, English, Mandarin, and Cantonese mid-sentence. Our team annotates this the way real users experience it.
Malaysian Mandarin and Cantonese carry distinct regional character. Standard Chinese annotation misses this entirely.
Xtecq is built in Malaysia, for the SEA market. →
Focused on the data work that makes AI systems better — across the languages that matter in Southeast Asia.
Human preference data for reinforcement learning — ranking AI responses, rating output quality, and flagging culturally inappropriate content across SEA languages.
Human evaluation of AI-generated content for accuracy, tone, helpfulness, and cultural appropriateness — in the languages your users actually speak.
Text, image, audio, and video annotation — plus localized data collection from real SEA users with the right demographics and linguistic profiles.
We start small so you can verify quality before scaling. No lock-in, no pressure.
Share your project scope, target languages, and quality expectations. No commitment needed.
We deliver a small pilot batch. You evaluate the quality yourself before deciding to proceed.
We incorporate your feedback and finalize annotation guidelines before full production.
Ongoing delivery with regular QA checks. You stay in control of pace and scope.
Native-speaker annotators complete initial tasks following your custom guidelines.
A second annotator reviews and validates the work, flagging inconsistencies.
A project lead spot-checks final batches against your agreed quality benchmarks before delivery.
Xtecq operates out of Kuala Lumpur — one of SEA's most linguistically diverse cities. Our annotators are local professionals who use these languages every day. That lived experience is what makes the difference in human feedback data quality.
Our annotators are Malaysian professionals — not crowd workers. They bring domain awareness and real cultural fluency to every task.
Malaysians naturally switch between languages mid-sentence. Our team annotates this the way real users experience it.
We're on the ground in Malaysia — accessible, responsive, and operationally aligned with your time zone.
Example RLHF Task
"Rank these two AI responses for a Malaysian user asking about home loan eligibility. Consider tone, accuracy, and local banking context."
Formal & informal variants
Primary language focus
Malaysian Chinese context
Strong regional coverage
Malaysian dialect nuances
Growing capability
Manglish & formal EN
Highest coverage
We're a focused, specialist team — not a generic annotation platform. Here's what that means in practice.
Honest answers before you reach out.
We're a services company. We don't have a self-serve platform — we work directly with clients to understand their project needs and deliver via a managed process. That means more customization, more communication, and more accountability than a generic platform.
We maintain a core team of trained annotators and can scale based on project requirements. We're transparent about our capacity — we'll tell you upfront whether your project volume is something we can handle, and at what timeline. We don't overpromise.
We'll complete a small sample of your actual task — typically 50 to 200 items depending on complexity — at no cost. This gives you real data to evaluate our quality and annotation style before you decide whether to proceed with a full project.
All annotators sign NDAs before working on any client project. Client data is used solely for the purposes of the contracted task and is not shared, retained, or repurposed. We're happy to sign your NDA before any pilot begins.
We work with most standard formats — JSON, JSONL, CSV, Excel, and can adapt to client tooling. For image and video annotation we support COCO, Pascal VOC, YOLO, and others. Let us know your stack and we'll confirm compatibility.
We can usually begin a pilot within a few business days of receiving your brief and sample data. Full project onboarding — including guideline finalization and team calibration — typically takes one to two weeks depending on project complexity.
Send us a brief description of what you need. We'll respond within one business day with honest feedback on whether we're the right fit — and a free pilot offer if we are.
Start with a free pilot. No commitment, no contract. Just real data you can evaluate yourself.
Xtecq
+601 0938 8503