Portfolio

Oskar Allerslev

Quantitative Infrastructure & Actuarial Engineering

MSc Actuarial Mathematics student at UCPH and Actuarial Software Developer at Festina Finance. I specialize in quantitative risk management (QRM), stochastic calculus, and machine learning. My focus is translating advanced mathematical theory into scalable, low-latency pricing engines and infrastructure using C++, Rust, and R.

Download CV (PDF) LinkedIn
Oskar Allerslev
Based in Copenhagen

01

Experience & Education

Industry roles and degrees, newest first.

  1. Actuarial Software Developer

    Jul 2026 – Present
    Festina Finance Work Now

    Role focusing on financial software, risk management engines, and scalable investment infrastructure.

  2. Student Actuary, Insurance Insights

    Nov 2024 – Jun 2026
    PFA Pension Work

    Developed internal actuarial packages (Lifepack), engineered automated DevOps ETL pipelines, and built quantitative stakeholder reporting tools.

  3. MSc Actuarial Mathematics

    Sep 2025 – Jun 2027 (exp.)
    University of Copenhagen Education Now

    Focus: Advanced probability, QRM, ML, financial econometrics.

  4. BSc Actuarial Mathematics

    Sep 2022 – Jun 2025
    University of Copenhagen Education

    Thesis: Hybrid insurance pricing framework.
    Highlight: Fire severity modeling using Extreme Value Theory (EVT).

02

Core Toolkit

What I build with, and the mathematics I build on.

{ }Languages

JavaR (Advanced/Package Dev)SQLC++PythonLaTeX

⌁Tools & Infrastructure

GitGradleAzure DevOpsVS CodeShinyTidyverse

∑Quantitative Math

Measure-theoretic probability

σ-algebras, filtrations, conditional expectation and martingales — the rigorous footing the rest is built on.

Brownian motion & Itô calculus

Quadratic variation, Itô’s formula and SDEs — from geometric Brownian motion (the paths at the top) to rough volatility.

Arbitrage-free pricing

Risk-neutral valuation under ℚ, change of measure via Girsanov, and Monte Carlo when no closed form exists.

Life insurance mathematics

Multi-state models and Thiele’s equation for reserves — the math behind Lifepack and the life & pension calculations I implement in Java.

Extreme value theory

Peaks-over-threshold with a generalised Pareto tail, used to model fire-claim severity.

Statistics & learning

Applied daily for pricing, forecasting and reporting.

GLMTime Series (Cointegration)Machine Learning

03

Featured Infrastructure

Open a project to read more and try the live demos.

A high-performance backend built in Rust using Axum. Fetches ticker data and calculates Historical Value at Risk (VaR) based on continuous returns and empirical quantiles.

View on GitHub
Query Control Panel
Target URI

Advanced derivative pricing framework based on fractional Brownian motions (fBM). Solves Stochastic Differential Equations (SDEs) to capture the roughness of electricity spot prices and financial markets.

View on GitHub
Quantitative Research Note
Click to expand mathematical proof
Rough Heston Asset SDE
dSt=μStdt+VtStdWt1dS_t = \mu S_t dt + \sqrt{V_t} S_t dW_t^1
Variance Process (Rough Volatility)
Vt=V0exp⁡(ηBH(t)−12η2t2H)V_t = V_0 \exp\left( \eta B_H(t) - \frac{1}{2} \eta^2 t^{2H} \right)
Cholesky Covariance Decomposition

Simulating the fractional Brownian motion BH(t)B_H(t) for H<0.5H < 0.5 requires careful consideration of its auto-covariance structure. The C++ engine performs an exact Cholesky decomposition of the dense covariance matrix Σ=LLT\Sigma = LL^T to generate valid rough volatility paths, ensuring statistical integrity.

Predictive machine learning pipeline forecasting highly volatile energy spot prices across 6 distinct markets. Optimized for RMSE utilizing specialized XGBoost models and domain-specific meteorological features.

RtidymodelsXGBoostMachine Learning
View on GitHub

Comprehensive R package for life insurance mathematics. Implements robust object-oriented programming (OOP), unit testing, and full roxygen2 documentation for production use.

R console
> library(lifepack)
> Lambda <- function(x) matrix(c(-0.1, 0.1, 0.05, -0.05), 2, 2)
> R <- function(x, mu) matrix(c(mu, 0, 0, mu), 2, 2)
> equiv_premium(0, 80, Lambda, R, dR, 0.05, 0.03, 100)
[1] 0.04512
View on GitHub CRAN

04

Beyond Code

Optimization outside the terminal.

30.09s

50m Breaststroke · Personal Best

Applying data-driven optimization outside the terminal. Elite sprint swimmer utilizing structured biohacking and sleep analytics for peak performance. Personal Best: 30.09s in 50m Breaststroke at the Danish Open.

05

LeetCode & GitHub

Practice and output, live.

GitHub Contributions

LeetCode Profile

Get in touch

Let’s talk numbers & code.

Connect on LinkedIn Download CV
Esc