I’m Arafath Hossain, a technical product leader for data and AI.
Experience
I started my career in product management and market research, then moved into data science. Over 10+ years I’ve worked across ML modeling, AI tool development and system design, always from the product side: how a solution fits the business process, not just how the model performs.
Most recently, at a Fortune 500 company, I took a vague request to answer “why” questions and designed a staged solution: detect what changed, find what caused it, simulate the options and recommend what to do, with the requirements and architecture to build it. I led the detection and causal modeling, shipped a detection and reporting layer that processes billions of records at enterprise scale, and delivered the causal proof of concept, aligning engineering, product, data science, business and design along the way.
I also built and owned an evaluation pipeline for AI agents, so teams could trust what the agents produce. Earlier, I built churn, recommendation and anomaly-detection models and led the teams that turned them into enterprise solutions.
What I bring to every project
Start with the question
Most data projects go wrong before any modelling starts. I spend the first stretch understanding who decides what, and break the problem down until the real question is clear.
See the whole system
Data science, engineering and analytics each tend to optimize their own piece. I design how the pieces fit into the real process, so nothing breaks at the hand-offs.
Align the people, not just the model
Projects usually stall between teams, not inside the model. I bring the leaders who own the decision, the data and the process onto one plan early, and keep them there.
Prove it before you scale it
I build working prototypes on real data and validate them honestly, with evaluation built into anything AI. A clear no-go early is a good outcome too.
A small team, shaped to the problem
I lead every engagement myself. When a build needs more hands, I bring in trusted specialists from my network, such as developers and economists, so the team fits the problem rather than the other way round.
See how an engagement runs →Have a problem worth untangling?
Book a free 30-minute call and tell me what you’re working on.
Book a 30-minute call