Machine Learning Data Scientist

Matthias Lalisse — AI Researcher/Engineer

CONTACT

INFORMATION Current City: Rochester, NY

Phone: 585-***-****

E-mail: ********.*******@*****.***

Homepage: http://pages.jh.edu/~mlaliss1

GitHub: github.com/MatthiasRLalisse

Updated: April, 2025

CURRENT ROLES

Research Grantee: Institute for New Economic Thinking Jan. 2023-Present

• Post-doctoral research using machine learning methods for political science projects in collaboration with Research Director Thomas Ferguson.

• Team combines expertise in economics, political science, machine learning, statistics.

• Built and maintain a database of financial and network data from the FEC and LDA, demographic data from the US Census.

• Systems built include entity linkage, record deduplication with foundation LMs-in-the- loop (text embedding for pairwise similarity in clustering donors), vote prediction using featurized donor, electoral, demographic variables.

• Representative publications at https://www.ineteconomics.org/research/experts/mLalisse

• LDA processing pipeline is open-sourced at https://github.com/MatthiasRLalisse/lobbylinks AI Engineering Contractor Dec 2021-Present

• Delivering AI/ML system to industry and academic clients.

• Project areas include systems in Natural Language Processing (LDA, LM-embedding based topic modeling, NER/parsing, text classification with LMs) and Computer Vi- sion (object recognition, coarse- and fine-grained classification, segmentation, bounding boxes; optical text detection/recognition (OCR), motion detection for energy-use opti- mization of online image analysis in inventory-tracking; image captioning/indexing/clustering. NLP supporting academic steams for text data cleaning (content extraction from SEC fil- ings using RegEx). LLMs for steered data annotation. General data analysis, statistical inference.

• Stack: Computer vision: Object detection/recognition (detectron2, huggingface, ViTs); Python, Pytorch, Tensorflow, Huggingface Transformers, Spacy, NLTK, Stanford NLP Pipeline, Scikit-Learn; newly, LangChain (LLMs, RAG).

• MLOps: NVIDIA Triton Inference server deployment of inference pipelines PHD RESEARCH

Ph.D. Thesis: Structure Assembly in Neural Networks My thesis examined models of knowledge representation in neural networks using classi- cal representational models like Holographic Reduced Representations (HRRs) & Tensor Product Representations (TPRs) coupled with deep learning. In addition to formal results relating to those architectures, I developed three models for knowledge base completion that perform at the state of the art on entity linkage within a neurosymbolic paradigm. Key terms: Computational semantics/computational cognitive science. Neurosymbolic computation. Models of semantic composition. Computational models of common sense inference applied to knowledge base completion. HRRs and TPRs. Harmonic Grammar. EDUCATION Johns Hopkins University, Baltimore, MD

Ph.D., Cognitive Science 2015 - 2021

• Dissertation: Structure Assembly in Knowledge Base Representation Sept. 2021 1 of 5

• Advisers: Paul Smolensky, Kyle Rawlins

• Areas of Study: compositional semantics, computational linguistics, knowledge repre- sentation

University of Oxford, Oxford, UK

M.Phil (with Distinction), Linguistics, Philology and Phonetics 2015

• Thesis Title: Distinguishing Intersective and Non-Intersective Adjectives in Composi- tional Distributional Semantics

• Adviser: Ash Asudeh

• Ertegun Scholar, Full Scholarship in a realm of the Humanities (Linguistics & Philology) McGill University, Montreal, QC

B.A./M.A. English, Minor in Political Science 2007 - 2011 Advisers: Peter Gibian & Monica Popescu

RESEARCH

PAPERS/ARTICLES

Gun money predicts congressional voting better than party alone INET article June 2022

Measuring the impact of campaign finance on congressional voting: A machine learning approach

INET Working Paper Series Feb. 2022

Sinema and Manchin Flush With Lobbyist Contributions as They Hold Up Biden Agenda Data For Progress Research Blog Oct. 2021

Scalable Knowledge Base Completion with Superposition Memories with Eric Rosen & Paul Smolensky

Workshop on Gradient Symbolic Computation, Baltimore, MD Sept 2019 Distributed neural encoding of binding to thematic roles with Paul Smolensky Poster presentation @ MACSIM 8, New York City April 2019 Augmenting Compositional Models for Knowledge Base Completion Using Gradient Rep- resentations. with Paul Smolensky

Proceedings of the Society for Computation in Linguistics (LSA), New York City Supervaluationist Semantics for Absolute Gradability. May 2017 CLS 53, University of Chicago

DATA PROJECTS

LobbyLinks

• A processing pipeline for extracting links between government entities (legislators, agen- cies...) and lobbyists operating on behalf of private organizations.

• Uses government filings and ML tools to process unstructured text data and produce visualizations (sample linked above) of lobbying networks.

• Designed to serve as a resource for investigating how lobbyist penetration of the legisla- tive process shapes policy.

TECHNICAL SKILLS

Programming languages: Expert: Python. Coversant: MATLAB, R. Machine learning libraries: Pytorch, Tensorflow, Scikit-Learn, OpenCV, ONNX ML problem domains: Proficient: Knowledge base completion. Conversant: Language modeling, Supervised and unsupervised clustering, ML for high-dimensional data (e.g. fMRI, campaign finance)

2 of 5

PRESENTATIONS

Distributed neural encoding of binding to thematic roles MACISM 8, NYU. April 2019

Augmenting Compositional Graph Completiom with Harmony Networks LSA-SCiL, New York City Jan 2019

Maximality, Minimality and Absoluteness in Delineation Semantics for Gradable Predicates MACSIM 7, Georgetown University. October 2017

TEACHING

EXPERIENCE Teaching Assistant for Foundations of Cognitive Science Spring 2018 Johns Hopkins University. Instructor: Paul Smolensky Teaching Assistant for Cognition Fall 2017

Johns Hopkins University. Instructor: Tal Linzen

Teaching Assistant for Phonology Spring 2017

Johns Hopkins University. Instructor: Colin Wilson Teaching Assistant for Language and Mind Fall 2016 Johns Hopkins University. Instructor: Julia Yarmolinskaya Teaching Assistant for Mathematical Models of Language Spring 2016 Johns Hopkins University. Instructor: Kyle Rawlins ESL Teacher 2012-2013

Centre de perfectionnement linguistique. Montreal, QC. Teaching Assistant for Postcolonial Literature 2010 McGill University. Instructor: Monica Popescu

GUEST LECTURES

Semantics, Pragmatics, & Discourse. October 2017

Johns Hopkins University. Course: Cognition, taught by Tal Linzen Lexical and Compositional Semantics. October 2016

Johns Hopkins University. Course: Language and Mind, taught by Julia Yarmolinskaya AWARDS AND

HONORS

University of Oxford

• Ertegun Scholarship ("full-ride") 2013-2015

• Faith Ivens-Franklin Travel Award Aug. 2014

• Linguistics Department Travel Grant Aug. 2014

McGill University

• Mary Keenan Scholarship in English 2010

• Provost’s Graduate Fellowship 2010

• Molson Chair Graduate Fellowship 2010

• Chester Macnaghten Prize in Creative Writing 2010

• Charles William Snyder Memorial Scholarship 2009

• Shakespeare Scholarship 2009

• Dean’s Honour List 2009-10

• Golden Key Honour Society 2009

3 of 5

VOLUNTEER

EXPERIENCE UNITE HERE, Atlanta, GA

Volunteer canvasser: Georgia Senate runoff Dec. 2020/Jan. 2021

• Canvassed in the Atlanta metro area to elect Rev. Raphael Warnock and Jon Ossoff to the US Senate.

AOC for Congress, remote from Rochester, NY

Data volunteer 2020-Present

• Analysis of text data using machine learning models such as BERT.

"Supervol Squad" member Summer/Fall 2020

• Organized and led trainings and phone banks to promote the Green New Deal and Alexandria Ocasio-Cortez during the 2020 election cycle. Teachers and Researchers United (SEIU Local 300), Baltimore, MD Worker-organizer 2015-Present

• Former AFT affiliate, current SEIU affiliate.

• Organized graduate workers of Johns Hopkins University around unionization, health- care, job security, and national policy issues such as immigration and taxation. Citizens Climate Lobby, Baltimore, MD

Citizen lobbyist 2015-2017

• Met with Congressional staff to advocate for CCL’s Carbon Fee and Dividend policy, a national carbon tax.

Keble College MCR, Oxford, UK

Environment Officer Oct. 2014-July 2015

• Organized green events for graduate students at Keble College, Oxford. Push Your Parents, Oxford, UK

Secretary Oct. 2014-June 2015

• Bottom-lined administrative functions for an advocacy group mobilizing students to persuade their parents to divest from fossil fuels. NON-ACADEMIC

EMPLOYMENT Indigo Bookstores, Montreal, QC

Book stocker and bookseller 2011-2013

• Stocked books in the in-store warehouse of a Canadian bookstore chain, and also worked on the sales floor.

IGA, Montreal, QC

Grocery store product sample distributor 2008-2010

• Worked through college distributing food samples in a Canadian grocery store chain. French Territories Translation, Rochester, NY

French-English Translator 2006-2012

• Translated documents including press releases, periodicals, employee and technical manuals, legal contracts, and essays for publication from French to English for busi- ness and academic clients.

CGI Communications, Rochester, NY

Script-writer Summer 2008

• Produced informational scripts for videos about cities and towns as part of CGI’s eLocalLink business.

4 of 5

RESEARCH

EMPLOYMENT McGill University, Montreal, QC

Research Assistant 2012-2013

Principal Investigator: Saleem Razack (Centre for Medical Education) Research Assistant Summer 2009, 2010

Professors Alanna Thain and Peter Gibian.

OTHER TRAINING NASSLLI, New Brunswick, NJ August 2016

• North American Summer School in Logic, Language and Information

• Rutgers University

ESSLLI, Tübingen, Germany August 2014

• European Summer School in Logic, Language and Information

• University of Tübingen

ACTL, London, UK Fall 2014

• Advanced Core Training in Linguistics

• University College London

CONFERENCE

ORGANIZATION The Humanities in the 21st Century March 2014

• Funded by The Oxford Research Centre for the Humanities (TORCH) and the Mica and Ahmet Ertegun Scholarship Programme.

5 of 5

Contact this candidate