Work
Personal projects and professional work — from open-source tooling to production marketplace systems.
Projects
Quoin
open sourceA developer workflow library for AI-assisted coding. Structured planning, implementation, and review skills — built so developers stay in control while leveraging Claude.
More projects
- Pricing Endogeneity PlaygroundA simulation showing how a well-fitting ML model recovers the wrong sign of price elasticity under endogeneity, while 2SLS gets it right. Companion code to "The ML Trap".GitHub · Read the essay →
- Firm Behavior under AIAn agent-based simulation of firms choosing human, AI-augmented, or automated production.
- AI tutor for kidsBuilt on OpenClaw; details when it's shareable.in progress
- Teaching materialsCourse notebooks on ML fundamentals and A/B testing, from my lecturing work.
- t-test labAn interactive Monte Carlo simulator that shows when Welch's t-test works, when it breaks, and what breaks it — skewness, heavy tails, and sample-size imbalance.
Experience
- 2022 – present
Careem (an Uber company) — Staff Data Scientist, Dubai
Experimentation platform tech lead. QUIK dynamic pricing, captain churn prevention, dispatch optimization, marketplace simulation. Promoted Senior → Staff in 2025.
- 2018 – 2022
MegaFon — Lead Data Scientist
Led a DS team: +10% subscriber inflow via retail-footprint modeling, +3–7% gross margin on device pricing, +15% effective ad spend via marketing-mix modeling, RL-based next-best-action.
- 2016 – 2018
Universitat Pompeu Fabra — Researcher, Barcelona
Network science and social-media research: large-scale data collection, filter-bubble measurement.
- 2012 – 2014
Otkritie Bank — Portfolio Manager, Retail Banking
Credit scoring, default modeling, exposure management — earliest risk-adjacent ML work.
- 2020 – 2021
Teaching — MegaFon academy, Mail.RU Technopark, SF Education
Intro ML, linear models, and experiment design (A/B testing, uplift modeling).