What Our Clients Say
Real feedback from organisations we have partnered with on AI quality assurance, adversarial testing, and monitoring engagements.
Back to HomeClient Testimonials
Lim Tze Wei
CTO, Kuala Lumpur
We brought prismars.pro in to set up a QA framework for our recommendation engine. They were thorough in understanding our architecture and built a test suite that caught issues we had been overlooking for months. The handoff documentation was detailed enough that our internal team picked it up within a week.
22 January 2026
Nur Aisyah Binti Razak
Head of Data, Penang
The adversarial robustness testing uncovered several input patterns that could have caused real problems in production. The severity-rated report made it straightforward to prioritise fixes. One area for improvement would be providing more context on remediation timelines, but overall the engagement was valuable.
5 February 2026
Kumar Subramaniam
VP Engineering, Cyberjaya
prismars.pro set up our entire AI monitoring stack — dashboards, alerts, drift detection — in about five weeks. The training sessions for our ops team were well-structured and practical. Six months on, we are still using the exact framework they delivered, with only minor adjustments on our side.
28 January 2026
Chan Yen Ling
Product Manager, Johor Bahru
We needed an external perspective on our NLP model before a major release. prismars.pro's QA program was exactly the right fit. They found edge cases in multilingual inputs that we had not anticipated, and the continuous evaluation pipeline they set up has been catching regressions since day one.
10 February 2026
Ahmad Hafiz
Lead ML Engineer, Kuala Lumpur
Good experience with the adversarial robustness testing. The team was professional and responsive. The final report had clear severity ratings and practical hardening steps. Took slightly longer than expected due to the complexity of our model, but the output quality justified the extra time.
1 February 2026
Priya Gopal
Director of AI, Petaling Jaya
We engaged prismars.pro for two services — QA and monitoring — across three production models. Their collaborative approach made it easy to integrate the work into our existing workflows. The documentation they delivered is now part of our standard engineering onboarding for new hires.
15 January 2026
Success Stories
Challenge
A fintech company in Kuala Lumpur had deployed a credit scoring model but had no structured testing around it. Edge cases in non-standard income profiles were causing inconsistent outputs, and the team had no way to catch regressions between model updates.
Solution
prismars.pro delivered the AI Quality Assurance Program over eight weeks. We built a comprehensive test suite covering 200+ scenarios including edge cases, integration tests, and automated regression checks integrated into their CI/CD pipeline.
Results
Regression detection improved from manual spot checks to automated catches within minutes of deployment. The team identified and resolved 14 previously unknown edge cases. Model update confidence increased significantly, reducing review cycles by roughly 40%.
Challenge
An e-commerce platform processing product images through a classification model noticed accuracy drops after seasonal catalogue changes. Without monitoring, these drops were only discovered weeks later through customer complaints.
Solution
prismars.pro implemented the AI Monitoring and Observability Setup over five weeks — including drift detection dashboards, feature stability monitoring, and an alert system calibrated to their specific accuracy thresholds.
Results
The team now detects accuracy drops within hours instead of weeks. Mean-time-to-detect fell from approximately 18 days to under 4 hours. The platform processes seasonal transitions with clear visibility into model behaviour, enabling proactive retraining decisions.
Challenge
A healthcare analytics company preparing to deploy a diagnostic support model needed independent validation of its robustness before clinical pilot. Internal testing was limited to standard benchmarks and did not cover adversarial scenarios.
Solution
prismars.pro conducted Adversarial Robustness Testing over four weeks, simulating input perturbations, noise injection, and out-of-distribution scenarios specific to medical imaging data. All findings were documented with severity ratings.
Results
Seven vulnerability patterns were identified, three rated high severity. The hardening recommendations were implemented before the clinical pilot began. The company was able to present the robustness report to their regulatory compliance team as part of the pilot approval documentation.
Trust Indicators
Years Active
Clients Served
Average Rating
Repeat Clients
Contact Information
Phone
+60 3-2171 6843
Office
No. 8, Persiaran Stonor, KL
Hours
Mon–Fri: 9AM–6PM
Ready to Join These Organisations?
If you are looking for a structured, transparent partner for AI quality — we would be glad to start the conversation.
Contact Our Team