Software Engineer Machine Learning

BECKY MASHAIDO Scholar Github Mail LinkedIn

********@********.***

Academics: Northeastern University

Master of Science (M.S.), Computer Science

University of California, Berkeley

Bachelor of Science (B.S.), Applied Mathematics

Google software engineer (3 years of non-internship experience) with a strong foundation in mathematics and focus on backend (APIs, servers, databases, cloud, distributed systems), machine learning and deep learning. Demonstrated expertise in achieving unique results for top tier companies and research institutions. Excels in identifying customer needs, designing, optimizing and scaling systems, developing critical business insights, leadership and mentoring other engineers. Frameworks and Technical Skills: Google (& Amazon) tech stack, Git version control, Keras, TensorFlow, PyTorch Programming Languages: Java, Python, C++, C, SQL, NoSQL, MATLAB, Microsoft Z3 Theorem Prover, SMT2 Solver MOST RECENT EXPERIENCE - SOFTWARE ENGINEERING

Google Sunnyvale, CA

Software Engineer 2022-2024

o Designed, built and maintained infrastructure, dashboards, applications, pipelines, internal software, distributed systems and feature requests that scaled performance by more than 15% and cut down organizational costs and time worth thousands of dollars

o Performed groundbreaking fleet optimization research techniques such as per-shard sorting for metadata, which cut down operational costs for Google Cloud by ~20%

o Engineered, tested, debugged, and deployed code to production o Identified and Prototyped high-opportunity projects that solved critical business needs o Mentored teammates by reviewing code and providing feedback to ensure best engineering practices o Minimized the time taken to production for my org by 60% through improving documentation on internal resources Amazon Palo Alto, CA

Software Development Engineer Intern Summer 2021

o Improved customer visibility by 100% and total revenue by $1B through an Application Migration and Modernization workflow o Designed and implemented asynchronous processes, APIs and infrastructure that extends the AWS serverless architecture for internal consumption

o Engineered an extension of the AWS distributed storage system that is scalable, fault-tolerant, low cost and easy to manage o Tested, debugged, built, and deployed code to production o Caught and fixed critical bugs for senior teammates, through code reviews o Enhanced project design workflows for Amazon interns, through design reviews VMware CodeHouse Palo Alto Palo Alto, CA

Software Engineer Hackathon Summer 2020

o Identified and led a team of six graduate-level students in creating a third-party application Tweet-Speak, that interfaces with the Twitter feed to provide direct audio access to tweets under user-specified hashtags o Won multiple category awards for improving audio accessibility for the blind and visually impaired Twitter users Northeastern University San Francisco, CA

Research Software Engineer 2019-2020

o Improved student grades with a 97% pass rate through implementing a mastery grading system that allowed students to get over common learning huddles

o Developed and launched APIs that increased grading efficiency of professors by 45% o Documented the new gradescope utility using sphinx, allowing professors visibility, easy access and adoption to their classes MOST RECENT EXPERIENCE - MACHINE LEARNING

Graduate Research Intern Matrix AI Consortium

Low-Precision Arithmetic in Deep Learning Fall 2021 o Conducted literature review on quantization and low-precision arithmetic o Performed inferences on the TinyML benchmark, in a 32-bit floating point numerical format o Published results in top ML conferences

Graduate Research Fellowship Northeastern University Detecting Neural Network Integrity Violations via Sensitive Samples Spring 2021 o Investigated the feasibility of rich methods from the field of symbolic reasoning to assess the vulnerability of deep neural networks against fault attacks

o Improved the efficiency of above symbolic methods such as SAT and SMT solvers by 25% o Applied the symbolic methods to networks that result from specific and well-known types of attacks such as Trojan attacks Graduate Research Assistant (MCADS Lab) Northeastern University Learning Grammar of Complex Activities via Deep Neural Networks Fall 2020 o Reduced computational complexity of weakly supervised action segmentation using a deep learning model that predicts the tasks of a video and actions happening in the video, under inaccessible frame-level label constraints o Analyzed model performance and implemented regularization techniques to reduce overfitting o Computed a quantitative score to measure the quality of attention for each test video and proposed further research directions Deep Learning Graduate Research Northeastern University Adversarial Training for Fairness in Deep Neural Networks Spring 2020 o Trained a deep neural network on the COMPAS dataset and analyzed the fairness of its predictions o Implemented adversarial learning which constrained the baseline model into making fairer recidivism predictions for Black inmates

Bold San Francisco, CA

Machine Learning Engineer - NLP 2018-2019

o Enhanced the baseline quality of semantic models using a semantically-ranked recommendation feature that increased user retention by 15 percent within 2 months

o Improved training time for supervised models by selecting useful features for input dataset based on the Levenshtein algorithm o Optimized search engine results by 15% using better-performing personalization and recommendation systems SELECTED SERVICES

Black High-School Girls Engineering Mentorship

o Launched personal initiative to raise the next generation of Black women in software engineering, computer vision and AI from underserved marginalized Black communities, globally She Codes Africa

o Designed and implemented learning modules for first-time African women in computing, to improve the representation of Black women in mathematics and computing careers

Relevant Courses: Computer Systems, Algorithms, Data Structures, Object-Oriented Design, Machine Learning, Linear Algebra, Deep Learning

Involvement: Google Engineers Code2040 CodePath /dev/color Black in AI She Codes Africa Women in Machine Learning Cal Women’s Rugby Afrotech

Awards: NCWIT Collegiate VMware Achieve Khoury Graduate Research AnitaB Grace Hopper Tapia P.E.O. AAUW Lime Melinda Gates Pivotal UC Berkeley Regents & Chancellor MasterCard Foundation Clinton Global Initiative Berkeley Big Ideas

PUBLICATIONS

Hamed et al. “ACTION: Automated Hardware-Software Codesign Framework for Low-precision Numerical FormaT SelectION in TinyML.” arxiv.org, 7 Jan 2022 [pdf].

Mashaido, Becky. “Learning Grammar of Complex Activities via Deep Neural Networks.” arxiv.org, 7 Jan 2021 [pdf]. Mashaido, Becky and Winston Tangongho. “A Tale of Fairness Revisited: Beyond Adversarial Learning for Deep Neural Network Fairness.” arxiv.org, 8 Jan 2021 [pdf].

Contact this candidate