Machine learning · Quantitative research
Iliass
Sijelmassi
Quantitative Researcher Intern at WorldQuant (Millennium) from October. Visiting Student Researcher at Stanford University. MSc Data Science & AI, École Polytechnique & HEC Paris.
00 · About
Machine learning researcher and engineer.
Currently at Stanford University, working on cardiac-event prediction from clinical waveform data with Prof. Louise Sun. Previously data science at Crédit Agricole CIB and software engineering at Infosys (Renault).
- Education
- MSc Data Science & AI — École Polytechnique & HEC Paris
- Now
- Visiting Student Researcher, Stanford University
- Base
- Stanford, CA / Paris, FR
01 · Work
Selected projects
Crypto Perpetuals Alpha
2026
Dollar-neutral long-short across 411 crypto perpetual futures. 379 candidate features pruned to 316 via Spearman IC, Ridge + XGBoost under a rolling 12-month walk-forward, with the cost analysis that kills the headline number.
#1 / 13 on OOS Sharpe · 9.75 gross @ lag 0 · 1.2 bps break-even cost
Python · XGBoost · Ridge · walk-forward CV
Cognitive Alpha
2025–2026
Per-pass decision quality on World Cup 2022 tracking data: trained Expected Threat, Spearman-style pitch control, and the gap between the pass played and the best option available.
126k possession actions · r = 0.93 vs reference xT · 5.4 GB tracking fused
Python · NumPy · Markov chains · Streamlit
Leukemia Survival Prediction
2025
Overall survival on clinical + genomic data from 24 hospitals, Cox proportional hazards. QRT / ENS national data challenge.
top 12% of the challenge · 4,500+ patients
Python · CoxPH · survival analysis
Options Market Maker
2025
Options market-making simulation: Black-Scholes quoting, inventory skew and delta hedging, evaluated over a 1,000-run backtest.
Python · FastAPI · Docker
Log Anomaly Detection
2023
LSTM sequence model learning the normal grammar of event logs, deployed behind Kafka streaming.
PyTorch · LSTM · Kafka
02 · Experience
Where I have worked
Quantitative Researcher Intern · WorldQuant (Millennium)
Systematic quantitative research within a Millennium systematic pod. Mission details confidential.
Visiting Student Researcher · Stanford University
Cardiac-event prediction under Dr. Louise Sun: deep learning on large-scale, noisy longitudinal clinical data, with leakage-controlled evaluation against clinical risk scores. First-author paper in preparation.
Data Scientist Intern · Crédit Agricole CIB
Audit-risk models on 200k+ records (XGBoost + regularized logistic regression), purged time-series validation, SHAP explainability for auditors.
Java Consultant Intern · Infosys — Renault
Backend modules for supply-chain platforms. Java, PostgreSQL, Oracle.
Software Developer Intern · Kuyper's Auto
Web interface and online reservation system, built directly with the owner.
- Languages
- Python · C++ · Java · SQL · R · TypeScript
- ML & Stats
- PyTorch · scikit-learn · XGBoost · survival analysis · time series
- Tools
- Docker · Git · MLflow · FastAPI · Kafka · Linux
03 · Research
Papers & reports
Predicting Severe Pericardial Tamponade After Cardiac Surgery
2026 · pdf ↗
MSc research paper · with Stanford University School of Medicine
Four model families on a Stanford surgical cohort (95 severe incident events, 6,761 controls). The calibrated pre-operative model reaches AUROC 0.74 and validates externally on MIMIC-IV; later ICU and bedside-waveform data add no established increment once monitoring-intensity confounds and selection optimism are controlled.
AUROC 0.74 internal · 0.695 external (MIMIC-IV) · 11,969 admissions
GMSK Modulation: Analysis and Implementation
2024 · pdf ↗
Technical report · signal processing
Gaussian Minimum Shift Keying: theoretical foundations, spectral efficiency analysis, and practical implementation considerations.
04 · Contact