Looking for a job

Alexandru Lecu

@alexandrulecu

Data scientist at Digital Science building production data systems and knowledge-graph-grounded LLM applications.

At Digital Science, I build and run production data systems for Dimensions, including a researcher profile curation API and journal data releases covering up to 10.8 million publications. I also own production pipelines for publications and grants, and have traced data-quality issues to root causes that informed team decisions.

At Technical University of Cluj-Napoca, I research knowledge-graph-grounded LLM systems for trustworthy AI. I designed and evaluated KG-augmented RAG for medical question answering, and built AlloyGraph, a multi-agent platform for superalloy property prediction and inverse design.

I also built BatteryKG, a provenance-aware knowledge graph that reconciles battery datasheet claims with measured degradation. My earlier work at Bosch Group Romania included predictive maintenance, anomaly detection and battery forecasting, as well as embedded and web software development.

Experience

Work history, roles, and key accomplishments

Current

PhD Researcher & Teaching Assistant

Current

Technical University of Cluj-Napoca

Oct 2022 - Present (4 years)

Designed and evaluated KG-augmented RAG for medical QA across eight LLMs and built LLM-based causal relation extraction into an OWL ontology. Developed and deployed AlloyGraph, a multi-agent platform for superalloy property prediction, and BatteryKG, a provenance-aware knowledge graph.

Current

Data Scientist

Current

Digital Science

Feb 2022 - Present (4 years 8 months)

Designed and deployed a production API for researcher profile curation and owned the Dimensions journals database, managing releases affecting up to 10.8M publications. Led data quality analyses and maintained production sync pipelines for publications and grants.

fastAPI BigQuery Kubernetes ArgoCD Sentry Python MongoDB PostgreSQL Luigi Jenkins GitHub Actions Gitlab CI Docker S3 GCP Pytest UV развертка Streamlit Claude Code

Data Scientist

Bosch Group Romania

Mar 2019 - Feb 2022 (2 years 11 months)

Built predictive maintenance models for industrial equipment and real-time anomaly detection for a printing line. Forecast battery charging needs and failure risk with regression and time-series models, and delivered dashboards and visual analytics for engineering teams.