Eric Sclafani

(631) 681-2797 · eric.sclafani321@gmail.com

I am a computational linguistics master's student at Stony Brook University working under Dr. Owen Rambow. My research primarily focuses on authorship attribution and how linguistic techniques can aid NLP systems in this task. My other research interests include finite-state morpho-phonology and sentiment analysis.

When I'm not working or teaching myself something, I have several hobbies. Playing video games is the main one. My favorite video games include: Terraria, Minecraft, Fallout: New Vegas, the Dark Souls franchise, Modern Warfare 2 (the old one), 20 Minutes Till Dawn.

Outside of video games, I also love to take long walks outdoors (especially in the cold), watching shows, and reading manga. My favorite type of media has to be lovecraftian horror, which in my opinion, is the most difficult to get right.


Education

Stony Brook University

Master's of Arts
Computational Linguistics

Bachelor's of Arts
Linguistics

Cumulative GPA: 3.7/4.0

August 2019 - May 2023

Suffolk County Community College

Associates of Arts
English

Cumulative GPA: 3.6/4

September 2017 - June 2019

Ward Melville High School

Regents diploma
September 2014 - June 2017

Projects

Gram2vec

Built a linguistic feature extractor in python under the supervision of Dr. Owen Rambow for a government-funded research project involving authorship attribution. This project will combine linguistic and NLP techniques in order to automatically detect the author of a document. Features of gram2vec include:
  • Vectorizes a collection of documents using spaCy and Sci-kit learn by searching for key grammatical and syntactic features indicative of author's stylistic tendencies.
  • Utilizing these stylistic vectors, applies a K-Nearest Neighbors classifier to perform ablation studies to tell which features are useful when identifying authors.
October 2022 - Present

Deltastar

Open source python package (in-progress) for creating deterministic finite-state transducers (FSTs). Began as a final project for Computational Linguistics 2 class and optimized thereafter. Features include:
  • Compiles string rewrite rules into FSTs which are applied to input strings to achieve a desired output
  • Constructs graphical representations of FSTs via the DOT graph description language

Click here to view the code!

April 2022 - Present

Multilingual Belief

Collaborated on a team headed by Dr. Owen Rambow on a research project utilizing the BERT multilingual transformer to model source-and-target belief in English, Mandarin, and Spanish natural language text. Responsibilities included:
  • Prepared input datasets using bash and submitted replicated model runs with varying initial seeds to a high performance GPU cluster over a remote SSH server
  • Wrote python script to batch calculate average and standard deviation from output JSON files to reduce the variance caused by making multiple model runs over separate seeds
  • Compiled cleaned data into spreadsheet for analysis
May 2022 - August 2022

Finite-state Morphological Analyzer for Tati

Worked with a team lead by Dr. Jeffrey Heinz to create a morphological analyzer for the understudied language Tati. This python program receives unconjugated verbs along with morphological tags and outputs a correctly conjugated Tati verb. Responsibilities included:
  • Leveraged Pynini, a finite-state machine (FSM) library, to construct a cascade of string rewrite transducers
  • Organized morphological verb data into spreadsheet
February 2021 - May 2022

Madlibs

Implementation of Mad Libs in python, complete with a fully-functional graphical user interface constructed from the Tkinter library. This was my first personal python project and was used as a code sample in my master's application. Features include:
  • Customized interface that prompts users for information to inject into the chosen Madlibs story

Click here to view the code!

November 2020 - April 2021

Experience

Graduate assistant

Institute for Advanced Computational Science
  • Managed front desk operations including kitchenette resupplying, unpacking deliveries, and taking inventory
  • Uploaded seminar recordings to Youtube and fulfilled conference room reservations
August 2021 - January 2022

Lot Associate

The Home Depot
  • Worked independently and with teammates to assist homeowners and contractors with loading purchases into vehicles
  • Served customers in a knowledgeable and friendly manner
November 2020 - August 2021

Lead Catering Bartender

Danford's Hotel and Marina
  • Responsible for bar functions including organization, setup, and stocking, either independently or with other bartenders
  • Using strong customer service skills, served customers standard and alcoholic drinks in a friendly and professional manner
  • Trained new bartenders on Danford’s operating procedures
November 2018 - February 2020

Skills

Programming Languages & Tools

Resume

Click here to view my resume