Jul 2020 – Dec 2020 · 5 mos · Bern, Berne, Switzerland

• I was a member of Swisscom's Applied AI and Analytics team, focused on applying NLP to automate internal business processes:
• Implemented and deployed a multilingual Transformer model (XLM-RoBERTa) to improve the routing accuracy of customer support tickets by >10% and generate a projected savings of >CHF 300k / year.
• Implemented and deployed a German named entity recognition system to automate the extraction of power-cut events from Swisscom energy provider documents.
• Established the team's first NLP/ML journal club, aimed at staying up to date with the current state-of-the-art and integrating it into Swisscom's services.
• Tech stack: PyTorch, HuggingFace Transformers, spaCy, Flask, Docker, Jenkins, Kubernetes
• I left Swisscom to care for my son during the COVID-19 pandemic.

Apr 2020 – May 2020 · 1 mo · Bern, Berne, Switzerland

• I developed and taught the graduate course "Practical Machine Learning for Physicists" at the Albert Einstein Centre for Fundamental Physics. The course involved two parts:
• Part I: An in-depth look at tree-based models such as random forests and gradient boosting.
• Part II: An introduction to topological data analysis with Mapper and persistent homology.
• A link to the course website can be found below.

Sep 2019 – May 2020 · 8 mos · Lausanne Metropolitan Area

• At L2F, I worked at the intersection of the product and research teams:
• GIOTTO. I led a team of 5 machine learning engineers to design and develop the company’s “no code” deep learning platform for computer vision tasks (https://giotto.ai/)
• TOPOLOGICAL MACHINE LEARNING. I implemented and evaluated topological algorithms for L2F's open source library giotto-tda to study data where the underlying structure is a non-Euclidean space.
• Some of my work has been published on Towards Data Science - see the links below!

Feb 2019 – Jul 2020 · 1 yr 5 mos · Greater Bern Area

• I co-developed and taught a recurring 6-month module on the foundations of data science and machine learning to industrial engineering students. Topics covered include the PyData stack, random forests, clustering, naive Bayes, model performance metrics, and cross-validation.

Sep 2017 – Jun 2019 · 1 yr 9 mos

• SPOUD is a Big Data startup focused on developing a platform that enables enterprises to locate, transport, and curate data assets. I led the company's data science efforts and worked on:
• ANOMALY DETECTION. Developed and implemented a novel unsupervised algorithm to automate the identification of failing components in a large IT application network. The research behind the algorithm was presented at Applied Machine Learning Days and the Swiss Conference on Data Science (see poster link below).
• APPLICATION PERFORMANCE MONITORING. Provided our customer's DevOps team with real-time performance metrics by developing and implementing Apache Flink pipelines that apply complex event processing to telemetry streams consisting of over 2,000 events/s.
• PREDICTIVE MAINTENANCE. Applied random forests and SVMs to identify factors influencing mean time between failure for over 200,000 units of industrial equipment to enable the customer to optimise their maintenance schedule effectively.

Sep 2013 – Jun 2017 · 3 yrs 9 mos

• My postdoctoral research was in the domain of theoretical particle physics, where I contributed towards furthering our understanding of:
• DARK MATTER. The nature of dark matter remains a deep puzzle in physics - we know it exists from astrophysical observations, but have yet to directly observe it interact with nuclear matter in Earthbound experiments. I proposed a novel framework to obtain limits on the mass of dark matter particles through high-precision calculations of quantities in nuclear physics.
• PHYSICS BEYOND THE STANDARD MODEL. The pressing question facing particle theorists today is: what lies beyond the Standard Model, the most successful theory to date? To help answer this question, I proposed a special class of theories where the Higgs boson emerges from the breaking of conformal symmetry.
• In addition to research, I was an active member of the Institute where I supervised students and gave a variety of lectures and seminars.

Jan 2012 – Jan 2013 · 1 yr · Greater Adelaide Area

• Developed major extension to the physics software package SOFTSUSY which uses an iterative algorithm to calculate the spectrum for a class of subatomic particles under active search at the Large Hadron Collider in CERN
• Writer of popular science articles