📄 cv

General Information

Full Name Shane van Heerden
Date of Birth 8 October
Languages English

Experience

  • Jul 2023 – Present
    Luno
    Senior Data Scientist (cf.)
    • Led key company project to develop real-time customer Anti-Money Laundering risk scores. Responsibilities included operationalising risk score models in Databricks using PySpark and MLFlow, engaging with Compliance stakeholders to construct risk indicators and advising on model parameter choices. Impact: Prevented potential regulatory fines up to R100 million.
    • Built declarative feature engineering infrastructure to service all ML models in Databricks. Impact: Around 10x reduction in feature creation and operationalisation time as well as improved scalability and data lineage alerting.
    • Won 2023 Lunaut of the Year in a company of 600+ people. (cf.)
  • Sep 2021 – Jun 2023
    Luno
    Data Scientist (cf.)
    • Led key company project for defining and tracking how customers move through various lifecycle statuses. Responsibilities included designing PySpark-based reverse ETL pipeline and monitoring reliability of upstream data dependencies. Impact: Critical component of a wider initiative to protecting $30 million in revenue and generating $1 million more.
    • Data Lead for CRM migration from Kustomer to Zendesk. Responsibilities included identifying critical data sources, coordinating with Engineers, and advising on new data architecture while considering cost and API rate limit constraints. Impact: Part of initiative to scale Customer Success operations to handle a customer-base of 100 million customers.
    • Led project to produce accurate inbound customer message forecasts for workforce hiring and scheduling optimisation. Involved exploring correlation between Bitcoin spending behaviour and messages volumes, and benchmarked predictive performance of ARIMA and Prophet forecasting models. Impact: 62% improvement in forecast accuracy.
    • Led project for Customer Success Workforce Team focussing to automating various productivity reports. Involved developing a PySpark-based data pipeline in Databricks, and displayed the productivity metrics to 20 Customer Success Team Leaders via Looker dashboards. Impact: Around 750 workforce hours saved per year.
    • Led a project tasked with identifying when the number of inbound customer conversations exceeded the available Customer Service capacity. Involved developing a set of indicator metrics and constructing a backlog monitoring dashboard using Looker. Impact: Better visibility and readiness for Customer Success to react to sudden demand spikes due to Bitcoin volatility.
    • Completed an advanced Data Science training program offered by Google's ML@CapitalG Team. Exposed to deep learning topics in Sequence Modelling and Reinforcement Learning.
  • Mar 2020 – Aug 2021
    Cape AI
    Machine Learning Engineer (cf.)
    • Led project for South African job recruitment firm, utilising advanced NLP to match resumes to job specs. Developed production-ready preprocessing pipelines on AWS EC2, stored/query results on Elasticsearch, fine-tuned BERT models, and served results via Streamlit. Impact: Client repositioned business offering around new AI-based recruitment matching system. (cf.)
    • Data Science Lead for an internal Cape AI venture aimed at connecting employees for knowledge-sharing opportunities by analysing unstructured text communications. Leveraged state-of-the-art NER techniques, built production-ready data pipelines interfacing with Neo4j database, and deployed and monitored ML model recommending possible connection opportunities. Impact: Venture secured external funding and remains operational. (cf.)
    • Led a project for a Netherlands-based electronic supply client which focused on using customer segmentation to discover high-value customers. This involved segmenting and clustering customer buying patterns by employing RFM analysis, and presenting results to the client via Streamlit. Impact: Identified 6 clients to target for re-activation who previously contributed ~€500k turnover. (cf.)
    • Led project for South African insurance provider to decrease administrative burden in handling customer queries. Analysed text communications and developed proof-of-concept chatbot using DialogFlow. Impact: Potential 60% containment rate in support ticket creation. (cf.)
    • Led project for client's data exchange platform, developing C# and Python-based ML solution to identify sensitive client data fields. Used Logistic Regression classifier and established active learning feedback loop between client's application and Azure cloud. Impact: 20x speed improvement in sensitive data field categorisation. (cf.)
  • Jan 2020 – Mar 2020
    DataProphet
    Data Science Intern (cf.)
    • Exposed to various data science-related topics and formed part of multiple teams responsible for installing DataProphet’s Prescribe product for their manufacturing clients. Projects involved discovering good operating regions for client's manufacturing machinery by employing fast Fourier transforms together with variational autoencoders using the Scikit-learn and Tensorflow Python packages.

Education

  • 2017 – 2019
    PhD (Data Science) (cf.)
    Stellenbosch University, Stellenbosch, South Africa
    • Research focused on the design and development of a Data Mining framework for quantifying and characterising road accident risk using machine learning and implemented as a Python-based solution.
  • 2013 – 2016
    BEng (Industrial Engineering) (cf.)
    Stellenbosch University, Stellenbosch, South Africa
    • Invited to the Golden Key International Honour Society. (cf.)
    • Won the prize for the best computer-based decision support system in the final year project. (cf.)

Academic exposure

  • Jun 2019
    European Conference on Operational Research (EURO)
    University College Dublin, Ireland
    • Presented research findings at the 30th EURO conference in Dublin, exposed to new ideas in the fields of Operations Research and Machine Learning. (cf.)
  • Sep 2018
    Deep Learning Indaba
    Stellenbosch University, South Africa
    • Attended a week-long meeting of the African Machine Learning community, exposed to teaching, research, exchange, and debate around the state-of-the-art in Machine Learning and Artificial Intelligence. (cf.)
  • 2016, '17, '18 & '19
    Operations Research Society of South Africa Conference
    South Africa
    • Presented research findings at the 2016, 2017, 2018 & 2019 ORSSA annual conferences.