Halil Erdogan
Work
  • Feb2025 - Present
    INFINIT
    Senior AI Engineer
    • Engineered LangChain-based multimodal agent for loan application processing, integrating text and video analysis to automate document verification and applicant interviews, reducing processing time by 80% and improving approval accuracy by 35%
    • Developed conversational AI voice agent using VAPI and OpenAI that reduced human dependency by 70% for routine inquiries while maintaining 95% customer satisfaction
    • Architected ML-based vehicle evaluation system reducing inventory review time by 50%, enabling efficient prioritization of high-value listings
  • Jan2019 - Sep2024
    Aiva Technologies
    Senior Machine Learning Engineer
    • Progressed from intern to technical team lead, driving growth to $1M ARR and securing major investment at $30M+ valuation
    • Architected high-performance ML pipeline processing 1M+ monthly compositions with 99.9% reliability and 90% reduced latency
    • Scaled AI music platform to 50,000+ MAU while maintaining 99.9% uptime; led migration from traditional ML to transformer-based models resulting in 300% YoY user growth
    • Designed and implemented fault-tolerant microservices architecture using Docker/Kubernetes, achieving 99.99% availability
    • Built real-time music generation APIs achieving 50ms response time and designed ETL infrastructure handling 100GB+ daily data
    • Developed custom analytics dashboard combining Grafana metrics and Python-based insights, improving decision-making time by 60%
    • Led technical due diligence process for successful fundraising rounds
    • Established comprehensive observability stack reducing MTTR by 75% using Prometheus/Grafana
  • Jun2018 - Sep2018
    Fraunhofer IAIS
    Research Intern
    • Engineered first-of-its-kind speech segmentation system reducing manual annotation time from hours to minutes, leading to widespread adoption across organizational projects
    • Collaborated with senior researchers to implement and evaluate novel audio processing algorithms for large-scale speech databases
    • Developed Python-based utilities for automated feature extraction and analysis of speech signals
    • Presented technical findings to cross-functional teams, receiving recognition for innovative approach to speech processing
  • Jun2017 - Sep2017
    Telecom ParisTech
    Research Intern
    • Developed novel probabilistic model for speech enhancement, published in ICASSP 2018 (Alpha-stable Low-rank Plus Residual Decomposition for Speech Enhancement)
    • Implemented advanced signal processing algorithms for noise reduction in audio recordings
    • Conducted extensive experimentation to validate performance of new speech enhancement techniques
    • Collaborated with research team to optimize model parameters and enhance real-world applicability
    • Contributed to open-source audio processing libraries, improving accessibility of research outcomes