Tehran, Iran · Sport Scientist · Applied ML Engineer
I work at the intersection of applied sport science, production machine learning, and elite combat-sport practice. As Algorithm Architect at Moneli Automation I built the bio-adaptive nutrition pipelines that power a deployed digital-health platform. As a national-team wrestler and coach, I have direct, longitudinal access to a population that academic data scientists rarely reach.
I sit at the intersection of three things that rarely come together in one person: applied sport-science training, hands-on production ML engineering, and direct longitudinal access to elite combat athletes.
Algorithm Architect at Moneli Automation, working directly under Dr. Esmalifalak. I architected the bio-adaptive nutrition algorithms — Python and scikit-learn pipelines that translate real-time physiological data into individualised, load-responsive recovery protocols, shipped to real users.
B.Sc. in Sport Sciences at the University of Tehran (GPA 17.28/20, ranked 1st in the department). Research assistant on a neuromechanical injury project; currently completing a review article on hormones, mental fatigue, and injury in elite athletes.
National-team wrestler and BJJ competitor for Iran. Three years coaching elite combat athletes — including World and Asian medallists — alongside Iran's Olympic wrestling head coach. The cohort most academic data scientists cannot reach.
I'm not someone who has read about applied sport-science ML. I have built it, shipped it, and seen what breaks when it meets a real population of users.
In a doctoral programme I would extend my applied work along three connected lines — bridging the methodological depth I'm still building with the elite-sport population I already have.
Combining wearable signals (HR, HRV, sleep), self-reported wellness, and training-load data into ML models that predict soft-tissue and overuse injury — and, critically, communicate their own uncertainty so coaches know when not to trust them.
Elite cohorts are tiny by ML standards. I want to use transfer learning from larger sub-elite datasets, synthetic data, and small-sample methods built around the practical reality of elite sport.
Building lightweight, interpretable interfaces that translate model output into something a strength-and-conditioning coach can actually use — and validating those tools in the field, not just on held-out data.
Short, self-contained pieces that demonstrate the methodology I want to take into doctoral work. Built on public or synthetic data so they are defensibly my own and reproducible end-to-end.
Applied ML · Wearable data · Uncertainty
Predicting next-day RPE from sleep, training load, and HRV — with split-conformal prediction intervals. The point predictions are usable; the intervals expose exactly the cross-athlete generalisation problem that motivates the research I want to do.
Signal processing · sEMG · Combat sport
A pure-Python sEMG pipeline — radix-2 FFT, Welch-style PSD, classical median / mean frequency features — plus a calibrated logistic classifier for fatigue detection during sustained isometric contractions. Beats the standard MDF-threshold rule by 10 percentage points on held-out athletes.
2022 — Present
Independent & various academies
2022 — 2024
University of Tehran
Private athletic club · 1 year
This elite combat-sport background gives me direct, longitudinal access to a population that is notoriously hard to recruit for sport-science research — a methodological asset I bring to any host lab.
A few moments from training, competition, and academic life.
B.Sc., Sport Sciences — GPA 17.28 / 20 (First-Class Honours equivalent)
Languages. Persian — Native · English — TOEFL iBT 89 (academic working proficiency).
I am open to doctoral opportunities, research collaborations, and conversations with labs working on athlete monitoring, fatigue, or injury prediction. The easiest way to reach me is email.