About
About.
Based in Islamabad. Working in Pakistan. Thinking about emerging markets.
Dr Mehak Rafiq leads the AI & Machine Learning Unit at Askari Bank in Islamabad, where a small in-house team has built the bank's production AI/ML capability from the ground up. She spent the previous decade as a tenured assistant professor and principal investigator at NUST, applying machine learning to cancer genomics and computational biology — twenty-two peer-reviewed publications, multiple research grants, and a PhD in Informatics from the University of Greenwich. The skills that find signal in messy biological data — pattern recognition, careful experimental design, an intolerance for unverifiable claims — turn out to be the same ones that build trustworthy AI in regulated banking.
Based in Islamabad. Working in Pakistan. Thinking about emerging markets.
Source: Google Scholar (as of May 2026)
Dr Mehak Rafiq
VP & Unit Head — AI & Machine Learning
In-house AI · Regulated banking · Emerging markets
Connect on LinkedIn →Risk
How I think about AI risk.
Banking AI is not consumer AI. The constraints are different and the failure modes are worse. My posture:
- Inspectability over capability. A model we cannot retrain is a model we cannot defend in a regulator's room. We use open-source models we can fine-tune in-house.
- Locality over latency. All inference runs on infrastructure we control. No data leaves the bank's network for an LLM call.
- Observability over uptime promises. Every production endpoint is instrumented with SigNoz. Failures are caught by us, not reported to us by users.
- Governance before scale. A draft AI governance charter — domains, ownership, escalation — sits behind every system before it goes live.