Experience

HT Ventures (January 2025–Present)

Role: AI Engineer (Remote)
Location: Hamburg, Germany

Architected and delivered an AI customer support copilot integrated with Gorgias + Shopify to streamline high-volume support operations.
Implemented reliable ticket-to-store matching and order-context retrieval so agents receive complete customer and order history instantly.
Built information extraction and response-suggestion workflows with proactive order-processing flaw detection to surface risks before they escalate.
Added intelligent escalation alerts for human agents, reducing average customer waiting time from ~14 days to near-immediate while exposing technical errors as soon as they occur.

Tools/Stack: Claude Code, MCP Servers, GCP, GitHub, PyCharm, GitHub Actions, Cloudflare, Redis, Celery, Open-source LLMs

Defined logistics requirements and computer vision solutions for railway transportation analytics.
Improved object detection performance by more than 2× through pipeline tuning, architecture updates, and dataset augmentation.
Engineered compact multi-class detection models for real-time inference and scalable deployment.
Implemented OCR for wagon and container numbers, with statistical validation under real-world conditions.

Collaborated with chemistry lab and R&D marketing teams to define data needs and project goals.
Built data modeling pipelines and automated workflows, integrating Power BI with Azure Databricks.
Developed LLM-based NLP solutions for sustainability analysis and review insight extraction.
Contributed to SalonLab with a secondary statistical model using NIR spectroscopy for hair damage identification.

Developed and maintained LLM-based retrieval-augmented generation (RAG) solutions.
Provided consulting on data science solution design for prospective clients.

Built neural machine translation and summarization for Swedish-to-English financial news, improving prediction turnaround time by 45%.
Led ETL pipeline development for SME financial data.

Worked on FaceNext: development and deployment of face-recognition access control across edge and cloud environments.
Built Street Visitor data engineering pipelines and predictive analysis from workers’ GPS device data.