Sabeti Lab
Jan 2022 ‐ present
The Sabeti Lab is a world-leading research group in the field of infectious diseases.
We are developing software for reconstructing how diseases spread during outbreaks,
and for mapping the genetic relationships between millions of viral genome sequences.
Partly, this involves bringing up research-quality prototype software to production
standards. Partly, this involves developing new approaches.
- Scaling of Bayesian phylogenetic tree inference methods. We aim to enable analyses of hundreds of genomes in near-real-time, and to make it possible to analyse millions of genomes.
- Production hardening of key computational engines and visualization of their inner workings and results (together with Fathom Information Design).
- Architectural advice and development to remove or mitigate scaling bottlenecks in data processing and visualization pipelines.
- Creation and/or identification of incipient analyses/techniques with high impact potential, followed by development and implementation at production scales.
Key technologies: C++, Java, Python, TypeScript.
Datadobi
Aug 2019 ‐ Dec 2021
Core development in Datadobi's flagship data migration product, DobiMigrate, which is used by
a large fraction of the Fortune 500 to efficiently and reliably migrate billions of files
across their data storage platforms. We have been instrumental in the development of the
following features:
- Maintaining a migration log, to certify the correctness of the content and metadata of every file.
- Designing and implementing a SQL-like DSL for aggregating massive hierarchical datasets, which powers DobiMigrate's advanced discovery features.
- Designing and implementing an internal library for naturally implementing large-scale cooperative
asynchronous processes, analogous to Java's
CompletableFutures. - Continuously improving the quality of DobiMigrate substantial code base (static nullability analyses, championing the use of immutable data structures, TDD, etc.)
Key technologies: Java, ANTLR, SQLite.
Corsearch
Nov 2018 ‐ Aug 2019
Built a trademark search engine, starting from a disparate set of R&D-quality data
science projects and ending in a coherent and production-ready service, with the
robustness, scalability and performance necessary to serve live client requests.
It can intelligently search through ~200 million trademark registrations worldwide within seconds.
Continuously interacted with data scientists, other front-end and back-end engineers,
product managers, architects and other stakeholders to shape its development.
Key technologies: Java, Spring Boot, Kafka, Kubernetes, Docker, SQL, MongoDB.
Sony
Jan 2017 ‐ Oct 2018
At Sony Techsoft, Sony's Belgian R&D center, led the development of
a back-end supporting storage and analysis of massive amounts
of sensor data, for release in an upcoming Sony product.
Sharply improved practices around clean architecture, refactoring and CI/CD,
with comprehensive testing suites, fully automated deployment to the cloud and
automated measurements of code quality.
Researched workings and possible uses of zero-knowledge proofs in the
context of smart contract blockchains (e.g., Ethereum).
Key technologies: Java, Python, AWS (including EC2, Lambda, DynamoDB, ECS and CloudFormation), Docker, Ansible, Jenkins, Ethereum.
Maastricht University
Oct 2016 ‐ Dec 2016
Built full stack of cBiT,
a web portal for exploring, searching through and retrieving research results
related to Biomaterial Transcriptomics. Organized as an intense agile project,
with weekly client feedback shaping the evolution of the product from
inception, through prototyping and onto production.
Key technologies: Angular2 + TypeScript (front-end), Python (back-end), ElasticSearch + PostgreSQL (databases), Ansible (deployment).
bPost
Oct 2016 ‐ Dec 2016
In conjunction with Data Minded,
analyzed a year's worth of operational data of the Belgian postal service, bPost,
to uncover insights for improving business processes. Interviewed a dozen key
people and mined large relevant data sets under intense time pressure to deliver
results. Our findings shaped the future direction of a high-profile initiative
at bPost.
Key technologies: HP Vertica (advanced SQL), Python (pandas, matplotlib).