engagements completed
average satisfaction rating
clients who return for further work
years of active consulting
Direct accounts from clients
"We'd been trying to get our ML deployment pipeline into a maintainable state for about eighteen months before engaging Cogniq. Within six weeks, the team had established a clear platform, written runbooks that our engineers actually found useful, and handed over everything in proper training sessions. The documentation quality alone was worth the engagement."
Ahmad Khairul
Head of Data Engineering · Financial services, KL
February 2026
"The annotation strategy engagement helped us fix problems we didn't fully know we had. The inter-annotator agreement benchmarks gave us an objective way to measure labeling quality for the first time — previously we were relying entirely on subjective review. The operations manual has been running our annotation programme for three months now without needing significant modification."
Priya Tharmalingam
ML Project Lead · Technology startup, Penang
January 2026
"The process mining engagement was genuinely surprising in terms of what it uncovered. We thought we understood our order fulfilment workflow reasonably well — the conformance analysis revealed three consistent deviation patterns we'd been unaware of, two of which had clear cost implications. The recommendations were practical and clearly prioritised. We've already actioned the top two."
Chong Siu Ming
Operations Director · Logistics company, Selangor
January 2026
"What I appreciated most was the straightforward communication throughout. If something was taking longer or running into complexity, Lian Wei would flag it early rather than us discovering it at the project close. The configured MLOps platform has been running our model updates reliably since the engagement closed — which is exactly what we needed."
Faridah Hassan
CTO · Insurance technology, KL
February 2026
"The annotation guideline document Cogniq produced for our computer vision project was noticeably more thorough than what we'd attempted internally. The tool configuration was also better than I expected — they'd thought through edge cases that hadn't occurred to us. The IAA benchmarks gave us something concrete to report to our ML team on data quality. Solid work, delivered on schedule."
Rajesh Nair
Data Science Manager · Manufacturing, Johor
January 2026
"We used Cogniq for a process mining engagement across our CRM event data. The process maps they produced made visible something we'd discussed abstractly for two years — where exactly leads were falling out of our pipeline and at what stages. The executive summary was presented to our board and informed a significant process redesign decision. We'll be engaging them again."
Sarah Wong
VP Sales Operations · Professional services, KL
February 2026
Selected engagement snapshots
Regional bank: unstable ML model updates
A Kuala Lumpur-based financial institution had three ML models in production but no consistent deployment process. Model updates were handled manually and inconsistently, leading to rollback incidents and monitoring gaps.
MLOps Setup engagement. Established a versioned deployment pipeline, configured monitoring dashboards with alert thresholds, authored operational runbooks, and ran three internal training sessions for the engineering team.
Zero unplanned rollbacks in the eight months following the engagement. Engineering team now handles model updates independently using the established process. Monitoring caught one performance drift issue proactively, four weeks before it would have affected outputs.
Logistics operator: supply chain process deviations
A Selangor logistics company suspected their goods-in process had inefficiencies but couldn't quantify the problem. Their ERP contained years of event log data that had never been systematically analysed.
AI Process Mining engagement. Extracted and structured seven years of ERP event logs, ran process model discovery, identified four deviation patterns in the goods-in workflow, and produced a prioritised optimisation report with business-language explanations.
Two of the four recommendations were implemented within two months of the report. The process redesign for one deviation pattern reduced average goods-in processing time by approximately 23%. The process maps continue to be used in operations review meetings.
Computer vision startup: annotation quality inconsistency
A Penang-based startup building an industrial defect detection system had grown their annotation workforce to twelve contractors but had no systematic quality framework. Inter-annotator agreement was unknown and model performance was inconsistent.
Data Annotation Strategy engagement. Developed annotation guidelines specific to industrial defect labeling, selected and configured an appropriate labeling tool, established IAA measurement protocols, and designed a review cycle that the team could operate independently.
IAA scores rose from an estimated 0.61 (measured retrospectively) to a consistent 0.84 within six weeks of implementing the new framework. Model performance on the test set improved measurably in the subsequent training run, attributed in part to improved data quality.
Reach us directly
Professional recognition
MDEC Registered Vendor
Malaysia Digital Economy Corporation approved AI services provider
PDPA Certified Practice
Personal Data Protection Act compliant data handling
Featured: AI in Asia 2025
Cited for MLOps implementation methodology and quality
Ready to explore what's possible?
There's no pressure in an initial conversation — just a chance to see if our approach is a good fit for your situation.
Start a Conversation