Marie Sklodowska-Curie Actions Doctoral Researcher
Misvik Biology Oy / MELOMANES EU Doctoral Network
Conducts doctoral research in cancer transcriptomics and nanomedicine safety while developing bioinformatics platforms for an EU-funded consortium.
Areas of work & evidence
Research & Cancer Transcriptomics Cancer transcriptomics, nanomedicine safety and patient-derived melanoma research.
- Designs and develops bioinformatics platforms supporting nanomedicine safety assessment and cancer research within the MELOMANES EU Doctoral Network.
- Applied transcriptomic and pathway-analysis methods to patient-derived melanoma models and nanoparticle-treatment responses.
- Delivered RNA-seq work with INSERM, Gustave Roussy and Cochin Institute in Paris; presented consortium updates and scientific workflow demonstrations.
Bioinformatics & RNA-seq Engineering Governed RNA-seq processing from data intake through analytical review and controlled downstream hand-off.
- Built RNA-seq workflows spanning dataset ingestion, row-identity resolution, gene-expression matrix canonicalization, normalization/QC review and controlled downstream hand-off.
- Integrated edgeR DGEList/filterByExpr/TMM/logCPM, PCA and limma MDS into browser-accessible workflows with explicit analytical state, candidate review and committed outputs.
- Developed modular R and Python pipelines for RNA-seq processing, statistical analysis, visualization and reuse of high-throughput transcriptomics data.
Scientific Platform Engineering Browser-accessible scientific systems connecting interfaces, orchestration, persistence and computational services.
- Connected PHP/JavaScript workspaces, Node.js orchestration, MariaDB persistence, R/Plumber scientific computation and Python/FastAPI research-synthesis services.
- Integrated R/Plumber statistical services with browser interfaces and backend orchestration to make analysis executable and inspectable by non-programming users.
- Developed modular scientific-software architecture spanning frontend workspaces, backend services, databases and computational runtimes.
AI & Inference Systems Local-LLM research infrastructure combining accelerated inference, bounded assistance and production resilience.
- Designed and production-validated a resilient GPU-accelerated inference architecture for ZAR AI using CUDA-enabled llama.cpp, Node-controlled GPU/CPU failover and automatic recovery, reducing representative inference from minutes to seconds while preserving an independent production CPU fallback.
- Built ZAR AI as a bounded scientific-assistance layer connecting deterministic tools, structured retrieval and locally hosted Mistral inference while keeping generated explanation separate from computed results.
Research Data & Governance Traceability, ownership and controlled analytical state across scientific workflows.
- Implemented metadata-driven provenance, checksum verification, structured issue reporting, private dataset ownership and review/commit workflows.
- Connected provenance, SHA-256 checksums, audit events, structured issue codes and owner-scoped datasets across scientific platform services.
- Designed analytical workflows around explicit candidate review, committed outputs and governed downstream hand-off.