Hermann Samimi

Machine Learning Engineer @FYUL

Tallinn, EE

Developed AI chatbots using Retrieval-Augmented Generation (RAG) and LangChain for real-time responses.•Built NLP modules for summarization, smart alerts, phishing detection, and threat analysis using containerized LLMs (Deepseek, Gemma3, Qwen3).•Implemented asynchronous anomaly detection pipelines integrating RabbitMQ, Logstash, Elasticsearch, Redis, Kibana, and Slack.•Created scalable Python services leveraging async workflows for high-performance data processing.•Applied ML models including Isolation Forest, LOF, ARIMA, Prophet, Autoencoders, and Bayesian Change Point Detection for anomaly detection.•Managed ML lifecycle using MLflow, Kubeflow, Hugging Face, GitLab CI/CD, and Docker/Nomad orchestration.•Engineered ML-driven security monitoring solutions detecting lateral movements, USB injections, insider misuse, and data exfiltration.•Deployed OSquery across 2,600 endpoints, streaming telemetry data into Elasticsearch for real-time security analytics.•Architected data ingestion pipelines from Cloudflare, AWS, Google Workspace, Okta, Jira, Confluence, and Bitbucket into Elasticsearch with real-time dashboards and alerts.•Designed optimized dimensional and normalized data schemas for analytics and high-performance queries.•Managed infrastructure-as-code (Terraform, Kubernetes), deploying containerized data services.•Built a scalable AWS S3-based data lake optimized with Parquet, Glue, and Athena for efficient querying.•Implemented data quality governance via Great Expectations in Airflow, and Avro schema versioning for Kafka.•Monitored and optimized ELK stack performance for sustained query efficiency and reliability.•Automated CI/CD pipelines for data engineering and ML workflows using GitLab.

Islamic Azad University

Associate's degree, Industrial Managing