Skip to content

Professional experience

Take a look at my LinkedIn profile!

Independent Contractor

Having worked in the industry for a few years and on a broad range of data science and machine learning projects, I have developed a strong understanding of the data science lifecycle and the challenges that come with it. I have accumulated experience in areas such as data engineering, data science, predictive analytics, machine learning engineering, software engineering, and systems architecture. I have also developed a strong appreciation for the importance of understanding the business context and how difficult it can be to effectively translate business problems into data science solutions.

As an independent contractor, I now help companies and teams to deliver value-driven solutions to their data science and machine learning problems. Take a look at the following sections to learn more about some of the projects I have worked on.

Reach out!

Nike (Ongoing…)

  • πŸ‘¨πŸ½β€πŸ­ Senior Machine Learning Engineer
  • πŸ‡³πŸ‡± Hilversum, NL
  • πŸ—“οΈ Aug 2022 - Present

Currently working with Nike as a Senior Machine Learning Engineer, where I develop, evaluate, and improve statistical models for long-term forecasting and demand sensing of Nike’s future consumer demand. I also develop and maintain the production pipelines that help us deliver these insights to ELT members and multiple planning teams around the globe on time and on a regular basis.

Beat (1yr 2mos)

  • πŸ‘¨πŸ½β€πŸ­ Machine Learning Software Engineer
  • πŸ‡³πŸ‡± Amsterdam, NL
  • πŸ—“οΈ Jun 2021 - Jul 2022

Beat is the fastest-growing ride-hailing service in Latin America. My mission was to enable and accelerate Beat’s sustainable growth by developing and applying Data Science and Machine Learning solutions to detect and prevent fraudulent actions, reduce financial losses and abuses, and ensure a safe environment for all users of our platform. All of this while keeping the business metrics healthy and the company growing!

πŸ”§ Feature Store Deployment (w/ Feast)

Led the development and implementation of a Feature Store solution to serve all ML and Analytics teams at Beat. Started by evaluating managed solutions such as Databricks, and SageMaker, conducting a two-week trial of Tecton, and finally settling for Feast as the registry and serving layers. Deployed the solution end-to-end on our Kubernetes cluster using our data-lake and Trino as the offline store, ElastiCache as the online store, and a low-latency gRPC Java service. Worked closely with Feast’s dev team by providing feedback and contributions to the open-source project. Currently working on standardising batch transformations in Fraud.

πŸ’³ Fraud Prevention

Led the productionisation of a fraud prevention system by (1) designing the solution’s architecture, (2) validating the assumptions and performance estimations of Data Scientists, (3) developing and deploying an ML (Argo) workflow that computes risk scores and pushes them to a Kafka topic, to be ingested to a fast-store (Aerospike) by a BE system, (4) configured monitoring for the infrastructure and application with Prometheus and Grafana, and (5) designing and running an online controlled experiment.

πŸš– Fraud Prediction

Worked with another Data Scientist on a fraud prediction problem where the labels were sparse, biased, noisy, and mostly positive and unlabelled (PU). We considered and explored several modeling options from naive binary classification with gradient boosting models to more robust active-learning approaches. The latest implementation lies somewhere in the middle, based on an iterative semi-supervised learning solution that yields twice as many fraud cases as the previous system while keeping precision extremely high (as per the pre-defined acceptance criteria).

Tiqets (2yrs 5mos)

  • πŸ‘¨πŸ½β€πŸ­ Machine Learning Engineer
  • πŸ‡³πŸ‡± Amsterdam, NL
  • πŸ—“οΈ Jan 2019 - May 2021

As a Machine Learning Engineer, I was part of Tiqets‘ core Data Team. I worked closely with data and business analysts, data engineers, as well as product owners and management team. I applied software development, data analytics, and machine learning to scale and operationalise statistical models and make the whole organisation more data-driven.

πŸ“ˆ Time Series Forecasting

Operationalised and automated sales forecasting at Tiqets by deploying a generalised ML workflow to run pre-processing, model selection, evaluation, and periodic updating of forecasted values for multiple and varying business contexts. Each task was distributed across an array of Celery workers deployed on our Kubernetes cluster. Forecasted values and model metadata was pushed to our DWH and visualised by business stakeholders using our BI tool. We used DataDogfor both application and infrastructure monitoring. For the most part, LSTM neural networks did not yield better results than classical approaches such as ARIMA and Exponential Smoothing, while an optimised Prophet model performed best for our type of data

Technologies: Kubernetes, ECR, S3, Redshift, SQLAlchemy, Alembic, DataDog, Celery, Redis, Dynaconf, FBProphet, mypy, pytests, scikit-learn, sphinx, statsmodels, pandas, numpy

πŸŽ₯ Recommender System

Improved recommendations across our platform by taking user-preference and item-similarity into account. Also, enriched popularity-based heuristics for cold-start instances by taking distance and recency into account. A curated version of an Amazon Personalize model now powers a great part of recommendations at Tiqets. To help the Data Team iterate faster and with greater confidence on new versions of the recommender models, we also implemented a time-dependent offline evaluation pipeline for recommender systems, curated for the e-commerce setting.

πŸ† Learning to Rank with Reinforcement Learning

To optimise any ranked list on our platform, I initially framed this task as a Supervised Machine-Learned Ranking problem, which involved comparing the predicted ranking to some ideal ranking using a metric such as nDCG and fitting a simple heuristic to this. To improve on this iteration, I implemented Bayesian Reinforcement Learning-to-Rank bandit strategies ( e.g. Thompson sampling ), which use explore and exploit to continuously learn and improve the rankings. As new data is collected through our event-pipeline, we use Airflow to frequently update item rankings on our platform (e.g. by re-sampling from an updated posterior distribution).

πŸ‘¨πŸ½β€πŸ’Ό Leadership / Management / Soft Skills

  • Helped management (CEO, COO, and CTO) and product owners in brainstorming sessions, defining company OKRs and team KPIs, and led meaningful reporting initiatives and important ad-hoc analysis.
  • The team grew from 3 to 15 people since I joined. I helped with the recruiting, assessing, and interviewing of candidates (including the current Head of Data) as well as organising and attending career fairs.
  • Provided guidance and supervision to two University students working on their Master’s Thesis. Both students produced valuable projects for Tiqets and finished with outstanding grades (8/10 and 8.5/10).
  • Integrated ideas from methodologies such as CRISP-DM to bring structure to the Data Science project lifecycle. Helped creating and prioritising the Data Science backlog, as well as making weekly meetings more fruitful and actionable.
  • Lead several company-wise trainings in general analytical competency and advanced BI (Looker) practice.

Accelogress (1yr 10mos)

  • πŸ‘¨πŸ½β€πŸ­ Software Engineer (Machine Learning)
  • πŸ‡¬πŸ‡§ Gildford, UK
  • πŸ—“οΈ Jun 2016 - Mar 2018

As a Machine Learning Engineer, I worked directly with the CEO and Lead Developer at Accelogress Ltd, a software consultancy company developing solutions using machine learning and API technologies. For the Save-a-Space project, I led the development of (1) the forecasting engine for car park availability using a robust machine-learning framework, and (2) a pre-processing layer and REST API to expose historical, realtime, and forecasted availability for multiple car parks around the UK, to our mobile app and web dashboards.

Technologies: Docker, nginx, Gunicorn, Django REST Framework, JavaScript, AngularJS, scikit-learn, MySQL

SEO (buzz)word-bank

Python | SQL | Docker | Big Data | Spark | AWS | Kubernetes | MLOps | CI/CD | Agile | Microservices