About Isroel Kogan
I’m an AI/ML engineer and consultant focused on building practical systems that help organizations forecast, make decisions, and use their data more effectively.
Through Kogan AI Consulting, I work with startups and growing companies that need experienced hands-on help with machine learning, optimization, AI search, and the systems that make those tools useful.
Discuss a project · View selected work
Professional Background
My background combines applied machine learning, decision science, and software engineering, with 8+ years of experience working across business and operational problems. I hold an M.A. in Mathematics and work across the full lifecycle: framing the problem, designing and evaluating models, implementing the solution, and explaining its tradeoffs to stakeholders.
At Thrasio, I worked as a Staff Data Scientist on pricing, forecasting, inventory, and supply-chain decisions. The work included developing production optimization systems that connected price response and demand to profitability, inventory constraints, and commercial objectives. An inventory-aware pricing initiative increased profitability by 15% while reducing excess inventory by 20%.
At Cadmus, my AI/ML engineering work includes service-level anomaly detection, data and model pipelines, monitoring, retraining workflows, and document retrieval for AI search. These projects put particular emphasis on operational reliability and meaningful evaluation, not just model development.
I have also developed an independent predictive modeling project using NYC housing complaint data. It addresses sparse-event prediction at large scale, with roughly 7 million previously generated risk scores evaluated against subsequent outcomes.
How I Think About AI/ML Work
A useful AI/ML system is not just a model. It needs a well-defined objective, sufficiently reliable data, evaluation tied to the actual use case, and a realistic path into a workflow people will use.
I try to start with the decision or problem, not a preferred technology. That might lead to forecasting, causal modeling, constrained optimization, a retrieval system, or a simpler approach. The choice should follow from the data, the tradeoffs, and what success would mean in practice.
I also believe limits should be visible. Calibration, uncertainty, weak data segments, and failure modes matter just as much as a headline metric — especially when predictions inform real decisions.
Why Work With Me
I work at the intersection of business objectives and technical implementation. That means I can help shape a problem, build the underlying system, evaluate whether it works, and communicate the result to both technical and non-technical stakeholders.
My experience spans distinct types of work: commercial decision systems, rare-event predictive modeling, AI document retrieval, and production monitoring infrastructure. The common skill is turning an ambiguous opportunity into something testable and useful.
I am comfortable working hands-on while also providing senior technical direction on architecture, validation, tradeoffs, and implementation priorities.
Kogan AI Consulting
Kogan AI Consulting is my founder-led consulting practice for teams that have meaningful data or knowledge assets and want to turn them into better decisions or working AI/ML products. Engagements can be project-based, advisory, or fractional technical leadership, depending on the need.
Have a Similar Problem?
If you’re evaluating an AI/ML opportunity, improving an existing system, or looking for someone to take a focused project from concept through implementation, I’d be glad to talk.