Blog
3D LiDAR Annotation Workflow: Lessons From an Internal Pilot
See how an internal pilot built a 3D LiDAR annotation workflow and exposed the quality controls needed before moving from KITTI to Waymo.
AI Council 2026: How Model Verification Shapes AI Quality
AI Council 2026 showed how verification shapes model training and agent safety. See what the sessions revealed about evaluation and runtime controls.
Careers in AI Aren’t Just for Engineers: Building the Workforce Behind AI Systems
AI projects don't scale because contributors appear when needed. They scale because hiring, qualification, and workforce planning begin long before the first task goes live.
Before You Scale AI Agents: 5 Lessons From Real Workflows
Five lessons from real AI agent workflows on defining system logic, building guardrails, and deciding where human review still matters.
RAG Pipelines Explained: How AI Finds and Uses Your Data
Follow the stages of a RAG pipeline from data preparation to context assembly and see how each step shapes the information an AI system receives.
Careers in AI Aren’t Just for Engineers: How Operations Teams Redesign AI Workflows
Before you automate, structure the work. See how system thinking and workflow design let teams scale without breaking, cutting hours to minutes, and growing 7x without adding headcount.
Inside LLM Systems: Designing for Risk & Reliability
An inside look at LLM systems, focusing on how system design, feedback loops, and observability influence risk and reliability over time.
Zero Rework at Scale: How We Stabilized Audio AI Production
Discover how Chemin stabilized audio AI production for gaming intelligence by embedding governance and multi-regional workflows to achieve a Zero Rework milestone.
Lessons from Building AI Agents in Client-Critical Workflows
See how a multilingual AI agent workflow cut manual handling time by 97% through staged automation and production controls.
Careers in AI Aren’t Just for Engineers: Why Delivery Is Central to Making AI Work
AI careers go far beyond coding. Explore how operations, quality control, and delivery teams transform AI into systems that work in the real world, and why these roles are driving the next wave of AI growth.
When AI Meets Agriculture: A New Way For Growers to See Their Orchards
A look at how Aerobotics uses AI and aerial imaging to give growers full-orchard visibility, changing how decisions are made across trees, blocks, and seasons.
What Real Malaysian Voices Reveal About Speech AI Limitations
Real voices from Melaka students exposed gaps in emotion recognition and pronunciation that research missed. Discover how regional accents, tone variation, and contextual ambiguity challenge mainstream Speech AI.
The New Sound of Gaming: How Audio AI Powers Reactive Player-Aware Worlds
Discover how studios are using Audio AI, speech diversity, and strong governance to build characters that react reliably to real players in real time.
The Anatomy of Listening: Designing Audio AI That Understands Context
The next frontier of Audio AI lies not in capturing clearer speech but in achieving deeper comprehension. Explore how embracing accent diversity and leveraging expert human feedback can pivot AI from hearing to comprehension.
Press Release: TDCX Group acquires SUPA to supercharge AI-enablement platform Chemin
Acquisition strengthens Chemin's capabilities in complex AI data services amid global demand surge.
Testing AI Hiring: What Reasoning Patterns Reveal About Bias in Southeast Asia
Would an AI model make the same hiring decision if the only thing that changed was the candidate's ethnicity?
Connect 4: The Illusion of Strategy in High-Performing LLMs
Given the recent interest in analyzing LLMs' reasoning abilities, we conducted tests on several LLM models by having them play Connect 4 to evaluate their reasoning ability.
AI Safety and Governance: Why It Matters More Than Ever
AI safety isn't just a technical issue; it's a societal one, affecting trust, governance, and the public good.
Chemin’s Advent of Code (Expanded, Curated, Verified) Dataset for Code Generation Evaluation and Training
Discover how Chemin's Advent of Code (Expanded, Curated, Verified) Dataset accelerates code generation research.
Strengthening AI Governance in Southeast Asia
Southeast Asia is rapidly adopting Generative AI (GenAI) and Large Language Models (LLMs), unlocking new business opportunities while introducing new risks.
Preparing Code Eval Datasets: Data Cleaning and Automated Code Execution for Advent of Code with Docker and Python
This blog outlines a system for processing Advent of Code submissions written in various languages.
DeepSeek R1 Crushes Advent of Code 2024: Our Latest Code Benchmark
Large Language Models (LLMs) for code generation have taken the software development world by storm, offering automated solutions to complex coding challenges.
Chemin's Bilingual Dataset for Evaluating Reasoning Skills in STEM Subjects
Fresh out of the oven! Our team just released a bilingual multimodal dataset for evaluating reasoning skills in STEM Subjects.
Press Release: TDCX and SUPA tie-up to help companies address a key barrier in generative AI adoption
Collaboration provides companies with a one-stop-solution for their data labeling needs.
Chemin Achieves SOC 2 Type II Certification: A New Level of Trust and Security for Our Customers
We are thrilled to announce that Chemin (formerly known as SUPA) has achieved SOC 2 Type II certification.
Press Release: Global Leader in AI Waste Intelligence Greyparrot.ai Expands to 89 Waste Categories, Powered by SUPA
Under every AI (Artificial Intelligence) model is a solid foundation of high-quality labeled data.
What Is Data Labeling? A Comprehensive Guide
Without properly labeled data, ML models struggle to understand key features, leading to unreliable results. This guide explores what data labeling is, how it works, different approaches, best practices, and its real-world applications.
What Happens When AI is Trained on Its Own Content?
Recent advances in generative AI image, text, and audio generation seem to promise endless potential. It's gotten to the point where researchers are exploring the use of synthetic data generated by AI to train next generation models.
GPT-4o-mini vs SEA-LIONv3: Bahasa Indonesia LLM Evaluation
A 50-task study tested Indonesian-specific prompts in English and Bahasa Indonesia to examine how model preferences varied across local-context tasks.
SEA-LIONv3 vs SahabatAI-v1: Bahasa Indonesia LLM Evaluation
SEA-LIONv3 and SahabatAI-v1 were tested on 50 Indonesian-specific tasks to examine how model preferences changed across task categories and prompt languages.
If You Want Data Annotation Done Well, Build The Right Workforce
Quality data annotation begins with building the right workforce. Building an effective, motivated workforce, however is no easy feat—especially when annotators are often crowdsourced & technically not full-time employees.