FinTech
- Building an agentic AI serving platform centered around observability and memory management
- Designing LLM tracing and observatory pipelines connected to knowledge management systems
- Developing multi-agent orchestration for financial automation workflows
- Evaluating and automating RAG pipelines and agent evaluation frameworks
Cloud-Native AI
- Fine-tuned open-source LLMs for domain-specific reasoning
- Designed distributed LLM training pipelines on multi-cloud environments
- Automated ML lifecycle on Kubernetes with custom operators
Automotive
- Built RAG applications for automotive consultation and technical documentation
- Developed custom parsing and preprocessing modules for diverse document formats
- Built data ingestion pipelines for multi-source automotive datasets
- Integrated corporate data across multiple European countries under GDPR constraints
- Graphiti — Knowledge graph construction for AI agents
- Serena — Semantic code retrieval and editing toolkit (MCP server)
- Aristotle — Blockchain-based artwork platform (own project)
Fine-tuning for domain adaptation · Human-in-the-loop agent design · Agent memory architectures · LLM observability and tracing Medium: @liebertar