Sofia Lawrie, PhD

Data Scientist

Greater Barcelona Metropolitan Area

About

Interested in practical approaches to problem solving and data-driven solutions. Always up for the challenge of learning, particularly new technologies.

Experience

  • Data Scientist at Gartner
    Apr 2024 - Present · 2 yrs 4 mos

    - Developed resume scoring models with high monetary impact (+2M US$ yearly), from experimental design of proof-of-concept to product deployment. Created automated reports to monitor scoring bias and fairness for legal compliance. - Designed and developed a highly scalable and configurable agentic application serving as a one-stop-shop for all HR related matters, using Streamlit for the UI, Fast API for the backend serving, Langgraph as an agent orchestrator, Milvus as a vector database and MLFlow for logging, monitoring and experimentation. - Expanded existing talent supply chain models to apply to new business cases, including the development of Azure ML pipelines and the maintenance of Power BI dashboard for end-user consumption. - Developed algorithms for information extraction from raw text into tabular format using both classical NLP and LLMs, and established LLM evaluation frameworks for benchmarking performance.

  • Data Scientist at Netquest
    Sep 2022 - Apr 2024 · 1 yr 8 mos

    - Performed data analysis to identify potential issues in complex transactional databases, quantify and report them. - Developed ETL pipelines for post-survey weighting, from proof of concept to production, in collaboration with data engineers - Optimized and automated company's internal processes through ensemble models, including an interpretability analysis using SHAP - Supported an external team developing an information extraction pipeline, including auditing and monitoring rule-based and machine learning-based NER (transformer) models. - Designed preprocessing pyspark pipelines to optimize storage of semi-structured data (achieved 30% reduction in storage size). - Migrated code to create experimental designs for choice-based conjoint studies from R to Python - Performed survival and churn analysis to extract online consumer panel KPIs

  • Universitat Pompeu Fabra (4 yrs 6 mos)
    • Research Data Scientist
      Oct 2017 - Mar 2022 · 4 yrs 6 mos

      - Implemented and trained neuronal network models for multivariate time series classification. - Designed and implemented pipelines for the analyses of behavioral and neuronal data recorded from monkeys. - Presented research results in oral and written formats. - Organized weekly group meetings. - Administrated research group's website.

    • Graduate Teaching Assistant
      Jan 2020 - Jul 2021 · 1 yr 7 mos

      Professor in Calculus II Seminars for Engineering Degrees, Department of Information and Communication Technologies

  • Undergraduate Teaching Assistant at Universidad Nacional de La Plata
    Apr 2016 - Aug 2016 · 5 mos

    Assistant in General Physics II Seminars, for Physics, Astronomy and Geophysics Degrees, Department of Physics.