Amin Belfkira

engineering student at CentraleSupélec

Download CV (PDF)

amin@belfkira.fr GitHub LinkedIn Paris, France

Experience

Quantitative Investment Strategies Structuring Intern

Société Générale Corporate and Investment Banking · Paris, France

Developed and backtested advanced quantitative equity-derivatives strategies focused on options, volatility and derivative signals. Improved the backtesting framework with Polars and Numba, and performed Greeks-based performance attribution across Delta, Vega and Gamma.

Python · Polars · Numba · Options · Derivatives

Data Scientist Intern

Crédit Agricole Corporate and Investment Bank · Paris, France

Researched neural-network approaches to P&L optimization for rates products using a one-factor Hull–White model. Developed NLP pipelines for financial signal extraction and fine-tuned an open-source SLM with TRL and qLoRA.

Python · NLP · Hugging Face · TRL · qLoRA · SLM

Education

Master of Engineering

CentraleSupélec — Université Paris-Saclay · Paris-Saclay, France

Statistics, high-performance computing, probability, big data, partial differential equations and project management. Ranked #1 worldwide in Mathematics (Shanghai Ranking 2022).

Master 2 High Performance Computing & Simulation (CHPS)

Université Paris-Saclay · Paris, France

Double degree pursued alongside the final year at CentraleSupélec. Advanced multicore programming, hardware accelerator programming (GPU), performance evaluation, advanced numerical methods, architecture and code optimization, and advanced compilation.

Academic Exchange

Oslo Business School — Oslo Metropolitan University · Oslo, Norway

Econometrics, derivatives and leadership.

Classe Préparatoire MPI

Lycée Aux Lazaristes · Lyon, France

Intensive coursework in mathematics, physics and computer science.

Selected projects

Physics-Informed Neural Networks for option pricing

Implemented a PINN to solve the Black–Scholes PDE for European option pricing through PDE-residual minimization and automatic differentiation.

Python · PINN · Black–Scholes · Deep Learning

Heston model calibration

Calibrated the Heston stochastic-volatility model with the Carr–Madan FFT method, then compared FFT and neural-network approaches for accuracy, speed and stability.

Python · Heston · FFT · Quant Finance

Product work

Skills

Languages
Python · C/C++ · SQL · OCaml · VBA
ML / Data
Hugging Face · TRL · qLoRA · Polars · Numba
Finance
Derivatives · Options · Stochastic models · Greeks
Tools
Git · Excel · Linux

Languages

  • French — Native
  • English — C1
  • German — B2
  • Norwegian — Basic

Interests

Football · Skiing · Piano · Hiking · Chess