Armandt Erasmus

Actuarial Analyst

Actuarial risk professional completing a dual BScHons in Physics and Financial Engineering, building scalable actuarial software and machine learning solutions to support risk management and data-driven decision-making.

Skills

  • Python
  • R (Shiny)
  • C++
  • SQL
  • PostgreSQL
  • MySQL
  • PL/SQL
  • MATLAB
  • SAS
  • Excel
  • VBA
  • NumPy
  • Pandas
  • Scikit-Learn
  • TensorFlow
  • PyTorch
  • PySpark
  • Databricks
  • Power BI
  • Django
  • Docker
  • Git
  • Azure
  • AWS

Selected Work

A car with front-end crash damage, the damaged region highlighted by Claimly

Claimly - Vehicle Damage Estimator

A full-stack app that takes a single photo of a damaged car and turns it into a priced, claim-ready repair estimate. It uses two YOLO segmentation models to identify the damaged parts, matches pricing to the car's make and model, and generates a PDF quote.

Django YOLO11 PyTorch Computer Vision
View project
All projects

Experience

  • Actuarial Analyst

    Feb 2026 - Present

    MiWay Insurance Limited

    • Led development of a proprietary claims estimation platform incorporating GLMs, Normalised Cost of Repair (NCR), salvage and claims development methodologies to support pricing, cost estimation and reserving, deployed on AWS and delivering R24M+ in annual cost savings.
    • Led the enterprise deployment of a document-forensics platform for insurance fraud detection, driving adoption across 3 business units and 75+ users, coordinating 14+ Manco stakeholders and replacing manual verification processes to deliver an estimated R7M annual benefit.
    • Architected and engineered a 24-module forensic fraud engine, developing the application layer in Django, data infrastructure in PostgreSQL, and integrating probabilistic evidence scoring and automated tamper detection to quantify document-level fraud risk and generate investigator-ready assessments within seconds.

  • Actuarial Analyst

    Jul 2025 - Feb 2026

    SolveCo Actuarial Services

    • Built a centralised multi-asset financial reporting engine (VBA) covering 50+ portfolios and 10+ asset classes, standardising valuations, fair-value movements, cash-flow timing, and FX attribution to deliver IFRS, US GAAP, and GIPS-aligned TWR/MWR performance reporting.
    • Developed a Python-based model validation framework that automated data preparation and testing for long-term insurance pricing models, improving governance and auditability while reducing testing timelines by 50%.

  • Machine Learning Engineer

    Jul 2023 - Jul 2024

    Google DSC

    • Led applied machine learning initiatives at the University of Pretoria, developing and delivering end-to-end ML solutions in Python, from data preprocessing and feature engineering through model training, evaluation and deployment using Scikit-learn and Streamlit for real-world business use cases.

Education

Certifications

  • Chartered Financial Analyst (CFA) Level I

    CFA Institute

  • Financial Modeling & Valuation Analyst (FMVA®)

    Corporate Finance Institute

  • Capital Markets & Securities Analyst (CMSA®)

    Corporate Finance Institute

  • IFRS 17: Insurance Contracts

    AICPA & CIMA

  • Data Science Professional Certificate

    IBM

  • Project Management Certificate

    Google