I'm a researcher, tech and research lead, mentor, developer, and innovator who can drive and guide product-focussed research projects from inception to conclusion.

Skills

PyTorch

0

Python

0

Haskell

0

Nix

0

Scala

0

SQL

0

Scientific Writing

0

Public Speaking

0

Teaching

0

Work Experience (11)

  • Developed research tracks and projects, oversaw and participated in their execution, and reported results to stakeholders and product teams
  • Liased with product teams to drive collaborative research agendas and align values
  • Worked on program synthesis in the context of human-machine interaction
  • Researched, designed, and built a state-of-the-art model for text-to-code translation in PyTorch
  • Developed a fast training method for grammar-constrained sequence decoding
  • As an NLP tech lead, conducted stand-ups, defined projects and goals, set expectations, and regularly liased with product teams and management
  • Mentored junior researchers and interns
  • Researched, designed, and implemented effective and novel methods for improving the time-quality-cost tradeoffs in data crowdsourcing
  • Researched, designed, and implemented unconventional and effective models for tree-structured decoding
  • Reimplemented state-of-the-art models for extractive document summarization in PyTorch
  • Reimplemented state-of-the-art models for document question-answering in PyTorch
  • Organized and hosted a weekly Haskell office hour and hacking group
  • Was an invited speaker at the Montreal NLP Meetup and gave a talk on closed-world factual question answering (see https://www.meetup.com/MTL-NLP/events/260045697/)
  • Reimplemented Unata's user-to-item and item-to-item recommender systems from the ground up in Scala and Spark
  • Maintained and refactored Unata's legacy recommender systems based on Postgres, Cassandra, and Redis
  • Hosted a 3.5 hour tutorial seminar on Bayesian machine learning at PyCon 2017 (a recording can be found at https://www.youtube.com/watch?v=fR5Wvb86-IU)
  • Leveraged functional programming and Akka to build various data ingestion and analysis pipelines for large data sets
  • Built a sentiment classification pipeline
  • Introduced a new mathematical framework for quantum coherent control interferometry
  • Studied quantum transport in large complex systems
  • Employed my own data analysis toolchain on the SciNet supercomputer cluster
  • Used OPENMP, MKL, MAGMA to speed up computation dramatically on modern multicore CPUs and NVIDIA Tesla GPUs with CUDA capability
  • Crunched through TBs of scientific data stored in HDF5 format
  • Dealt with correlated random variables with heavy-tailed marginal probability distributions
  • Employed random matrix theory to classify, interpret, and replicate numerical results
  • Studied quantum transport in complex systems
  • Performed large-scale Monte-Carlo simulations using a modularized software package written by myself in C
  • Developed my very own data analysis software stack
  • Implemented a toolchain in C, Python, Cython, and Mathematica for the creation and interpretation of multidimensional histograms
  • Deployed genetic optimization algorithms (CMA-ES) to find optimal configurations of complex, fully-connected networks
  • Used MPI-2 to delegate data streams from worker nodes to masters
  • Worked with the Black Forrest Grid supercomputer cluster
  • Graded various graduate courses in special topics of quantum mechanics
  • Hosted and tought in course exercise groups
  • Discovered a new way of entanglement detection through quantum interference
  • Optimized entanglement witnesses numerically in Mathematica
  • Graded various graduate courses in special topics of quantum mechanics
  • Hosted and tought in course exercise groups
  • Planned, implemented, deployed, and administered heterogeneous general-purpose computer lab
  • Setup and maintained both Linux and Windows workstations
  • Setup and administered Linux, NetWare, and OES servers
0

Projects (1)

Hasktorch
Aug 2019 - Current
  • Haskell
  • Torch
  • neural networks
  • generics
A Haskell library for tensor math and differentiable functional programming
  • Member of the core contributor team
  • Contributed massively to the typed tensor API and the Nix toolchain
  • Made it possible to check tensor shapes, data types, and compute devices at compile time
  • Implemented typed neural network layers including Transformers, LSTMs, and GRUs
  • Created typed neural network examples
  • Gave a live-coding demo of Hasktorch at the FP Berlin Meetup (a recording can be found at https://www.youtube.com/watch?v=ZnYa99QoznE&t=1689)

Volunteer

Jun 2020 - Aug 2020
Google Summer of Code Mentor
Haskell.org
Mentored a GSoC student in the "Custom Dataloader for Hasktorch" project

Education (2)

2008 - 2011
Doctor of Philosophy (Ph.D.)
 Theoretical and Mathematical Physics
The University of Freiburg
0
2002 - 2008
Diplom (German equivalent of M.S. degree)
 Theoretical and Mathematical Physics
University of Bayreuth
0

Languages

English

Fluent

German

Native Speaker

French

Basic Knowledge

Interests

Biking

Running

Open Source