John Rahmon

Engineered large-scale ETL pipelines migrating 4M+ customer accounts in Switzerland’s largest bank merger. Improved data workflow performance by 60% through SQL optimization and legacy system refactoring. Resolved 400+ critical migration issues ensuring reproducibility and reliability of financial data. Delivered scalable reporting dashboards for Account Managers supporting oversight of 3 key banking products. Designed and implemented data and AI infrastructure for a job-recommendation platform. Directed end-to-end ML/AI data pipelines processing 50K+ job ads monthly into structured NLP insights. Fine-tuned LLMs on a 3.1B token corpus using PyTorch FSDP on Azure ML, cutting training time by 20%. Built semantic modeling systems combining contrastive loss and BERT embeddings, achieving 35% less data needed to hit accuracy targets. Automated LLM training orchestration with Airflow, reducing manual developer effort by 45%. Deployed production-grade AI services with Terraform and CI/CD (G

Owned product vision and backlog for an AI-powered dental diagnostics platform, delivering a market-validated MVP that secured CHF 6M in funding. Led design of multimodal AI pipelines processing text and diagnostic images, migrating over 6 TB data to Azure to enable large-scale ML/NLP use. Built LLM-powered treatment recommendation chatbot indexing over 120,000 research publications. Developed NLP fact extraction systems automating structured knowledge ingestion. Achieved 20% model accuracy gain by designing novel deep learning contrastive loss functions for class-imbalanced data.