Halil Erdogan

CV

This page includes my experience timeline. Download the PDF version.

  • Feb 2025 - Present
    INFINIT
    Senior AI Engineer
    • Built an automated risk evaluation platform for credit underwriting, moving teams from ~4 hours of manual analysis per case to fully agentic analysis plus a ~3 minute manual verification step
    • Designed a multi-agent risk monitoring system that continuously evaluates portfolio exposure, flags anomalies early, and automates follow-up actions to reduce default and churn risk through faster intervention
    • Implemented an end-to-end contract automation system that generates documents faster and automatically verifies key values and constraints before submission
  • Jan 2019 - Sep 2024
    Aiva Technologies
    Senior Machine Learning Engineer
    • Progressed from intern to technical team lead in 3 years, supporting growth to $1M ARR and helping with technical due diligence during fundraising
    • Architected a high-performance ML pipeline for 1M+ monthly compositions with 99.9% reliability and 10x faster-than-real-time generation
    • Scaled the AI music platform to 50,000+ MAU, helping drive the transition from traditional ML systems to transformer and agentic architectures
  • Jun 2018 - Sep 2018
    Fraunhofer IAIS
    Research Intern
    • Built speech segmentation tooling that reduced manual annotation effort for large datasets
    • Collaborated with researchers to implement and evaluate audio processing methods for speech corpora
    • Developed Python utilities for feature extraction and signal analysis
    • Presented results and implementation details to research and engineering stakeholders
  • Jun 2017 - Sep 2017
    Telecom ParisTech
    Research Intern
    • Contributed to a probabilistic speech enhancement model published at ICASSP 2018
    • Implemented signal processing algorithms for noise reduction in challenging audio recordings
    • Ran experiments to evaluate model behavior across multiple datasets and recording conditions
    • Worked with the research team on parameter tuning and practical evaluation
    • Contributed code and documentation to reusable audio processing tooling