Dr Mehak Rafiq, PhD
VP & Unit Head — AI & Machine Learning at Askari Bank. Building production GenAI in regulated banking, in emerging markets.
The case
Most banks in emerging markets are buying AI the way they bought core banking in the 2000s — a multi-year contract, a seven-figure license, and a delivery date that slips for two years. It does not work. Traditional vendor deployment cycles in emerging market banking often stall at implementation.
I run the alternative. A small team, hired locally and trained on the actual stack. Open-source models on infrastructure we control. Code that lives in our own GitLab, behind our own firewall, instrumented with our own observability. Systems built and shipped for less than the markup on a single vendor PoC — and operated by the same engineers who built them.
The unfashionable claim: in regulated banking, in emerging markets, in-house is not a compromise. It is the only model that ships.
What I stand for
In-house
Code, models, infrastructure
Production
Live in regulated banking
Open-source
Models we can inspect and retrain
Self-hosted
Inference, observability, version control
Selected work
A working portfolio, not a résumé. Production AI from inside the bank, alongside peer-reviewed research from the lab.
voice banking for the mobile app
Voice-driven fund transfers, balance inquiries, bill payments, and statements. English and Urdu. Built in-house on FastAPI, Whisper, FunctionGemma, and Docker. Built proprietary agentic solutions that offset typical six-figure enterprise SaaS licensing fees.
Compliance
Continuous adverse-news screening across multiple sources, with semantic scoring used daily by the Compliance team for KYC and AML triage.
Chimera + AutoDock Vina, written for non-bioinformaticians
A practical tutorial taking life-science researchers from raw structure files to docked complexes without prior bioinformatics training. Cited 290 times — the most-cited paper of the lab and a primary teaching reference for early-career researchers entering computational structural biology.
JMIR Bioinformatics & Biotechnology · 2020 · 290 citations
Pro-apoptotic potential, computationally and in vitro
A combined computational and wet-lab evaluation of taxifolin's pro-apoptotic potential against hepatocellular carcinoma. Network analysis identified candidate targets; downstream assays validated the strongest hits. Representative of the lab's translational-bioinformatics output during the NUST years.
PeerJ · 2021 · 36 citations
About
The skills that find signal in messy biological data turn out to be the same ones that build trustworthy AI in regulated banking.
Working with me
I take a small number of advisory engagements outside Askari Bank. The fastest path is email.
Three lanes I prioritize →
VP & Unit Head — AI & Machine Learning
In-house AI · Regulated banking · Emerging markets
Connect on LinkedIn →