// CLIENT FEEDBACK
What clients say
about working with us.
First-hand accounts from Malaysian organisations that have completed engagements with Myvaltrix.
← Back to HomepageWhat clients have shared
"The knowledge graph they built has changed how our research team navigates our own archives. Things that used to require a manual search across three systems now surface automatically. The handover documentation was clear enough that our in-house engineer could maintain it without going back to Myvaltrix."
Nurul Zahirah
Head of Research, Kuala Lumpur
March 2025
"We commissioned the fraud analysis engagement after seeing a rise in unusual transaction patterns we couldn't explain with our existing rules. The model Myvaltrix developed gave us detection capability we'd been trying to build internally for two years. The false-positive rate was managed well — something our operations team had been worried about."
Chan Wei Ming
Risk Manager, Selangor
February 2025
"The ethics review was more substantive than I anticipated for a four-week engagement. Myvaltrix went through our customer recommendation system carefully and flagged several places where demographic proxies were influencing outputs in ways we hadn't intended. The recommendations were practical, not just theoretical observations."
Siti Rahimah
Product Lead, Kuala Lumpur
January 2025
"What I appreciated was that during the scoping call, Reza was honest about what a knowledge graph wouldn't solve for us. We ended up adjusting our expectations before the engagement started, which meant the actual work was targeted and useful rather than over-promised and disappointing."
Faridah Hassan
Data Strategy Director, Petaling Jaya
December 2024
"The milestone check-ins during the fraud analysis engagement were more useful than I'd expected them to be. We caught a data quality issue in week three that would have affected the model significantly. Having those structured touchpoints meant the problem surfaced early enough to fix."
Lim Kah Yin
Head of Analytics, Penang
November 2024
"We had previous experience with a larger consultancy where the work was done by a team we never met. Myvaltrix was different — we worked directly with Yee Lian throughout the ethics review, and she was accessible when we had questions between sessions. The report was dense with substance, not filler."
Ahmad Tarmizi
CTO, Shah Alam
October 2024
How engagements played out
// CASE STUDY 01 · KNOWLEDGE GRAPH · 10 WEEKS
Research institute with fragmented knowledge assets
CHALLENGE
A research institute had accumulated over twelve years of reports, papers, and datasets stored across separate systems with no shared indexing. Staff were rediscovering existing findings by accident, and the cost of that inefficiency was significant.
SOLUTION
Myvaltrix built an entity-relationship extraction pipeline that processed the institute's text corpus and populated a Neo4j graph database. A set of discovery queries was developed, enabling researchers to find related work by concept rather than keyword.
OUTCOME
Research discovery time reduced by approximately 60% for known-item searches. Three previously disconnected research threads were identified as overlapping within two weeks of the system going live. Internal adoption was high due to the quality of the handover documentation.
"We found connections in our own data that we'd been missing for years. That alone made the engagement worthwhile."
— Head of Research
// CASE STUDY 02 · FRAUD PATTERN ANALYSIS · 8 WEEKS
E-commerce platform with rising irregular activity
CHALLENGE
An e-commerce business was experiencing a notable increase in suspected fraudulent transactions that its existing rule-based filters weren't catching. Manual review was consuming significant operations time with inconsistent results.
SOLUTION
Myvaltrix developed a behavioural pattern detection model trained on 18 months of transaction history. Feature engineering surfaced time-of-day, session behaviour, and device consistency signals that the rule-based system had no visibility into. Thresholds were calibrated with the operations team to keep manual review volume manageable.
OUTCOME
Suspicious transaction flagging improved by 73% compared to the prior rule-based system. False-positive rate stayed within an agreed operational ceiling. Manual review volume dropped by roughly 40% within the first month of deployment.
"The threshold calibration work was where Myvaltrix's practical experience showed. They understood that a model that flags everything isn't useful."
— Risk Manager
// CASE STUDY 03 · AI ETHICS REVIEW · 4 WEEKS
Fintech deploying a customer credit scoring model
CHALLENGE
A fintech company was preparing to deploy a credit scoring model for loan eligibility decisions. Concerns had been raised internally about whether the model was treating different demographic groups fairly — but no one had the specific expertise to evaluate it systematically.
SOLUTION
Myvaltrix conducted a structured fairness analysis of the model's outputs across income brackets, age groups, and geographic regions. The review also examined the training data for historical bias and assessed the transparency of the scoring logic for customers who might dispute a decision.
OUTCOME
Two substantive fairness issues were identified and documented with severity ratings. The company addressed both before launch and used the report as part of their regulatory readiness documentation. Deployment proceeded approximately six weeks later than originally planned — a trade-off the company considered worthwhile.
"The review delayed our launch, but we launched with confidence. That mattered more than the timeline."
— CTO
ADDRESS
9 Lorong Kuda
50460 Kuala Lumpur, Malaysia