# Matthew Jeng > Software Engineer at Facebook Location: San Francisco, California, United States Profile: https://flows.cv/matthewjeng Experienced Backend/Full-Stack Developer. Skilled in Java, Go, Python, C, Javascript as well as with tools such as Postgres, Docker, Git, GDB. Strong engineering professional with a Bachelor’s Degree focused in Computer Science from UC Berkeley. ## Work Experience ### Software Engineer @ Facebook Jan 2020 – Present | Menlo Park, California, United States ### Computer Security Undergraduate Student Instructor @ University of California, Berkeley Jan 2019 – Jan 2019 | San Francisco Bay Area For a semester, I was a TA for the Computer Security class at Berkeley (https://cs161.org). This involved running weekly a discussion section, my own office hours, and generally being available - Piazza, in person - to help students with their homeworks, projects, and other questions they might have on class material. We cover topics such as (but not limited to): - Software vulnerabilities (taught primarily in C) and how to exploit them - Software defenses (ASLR, NX, canaries, memory safe languages) - Cryptography (block ciphers, encryption schemes, public key, signatures) - Network security (TCP/TLS, DNS/DNSSEC, SQL injections) - Web security (CSRF, XSS, cookies, phishing) - Miscellaneous security topics ### Software Engineer Intern @ Facebook Jan 2019 – Jan 2019 | Seattle, WA My primary project at Facebook involved working with Marketplace image enhancement code. I first had to generalize Marketplace image enhancement code in order to work for more objects. I modified code on the objects themselves, the code that the frontend used to query for these objects, as well as the image enhancement code itself. After the generalization, I backfilled the ~40 million objects newly eligible for enhancement, and ran an A/B test to confirm the new code creating and serving enhanced photos was generating only neutral or positive results. I finished my project early, and so had time to work on a few other tasks. The significant one out of these was a mini-project to add additional logging to Marketplace's composer. I added frontend click logging for when users selected certain options on the composer for mobile (iOS and Android) and on web, and added fields to existing backend logging after users submitted their items to Marketplace. This new data is used in Marketplace metric analysis. ### Full Stack Project Architect & Manager @ Pantheon Platform Jan 2019 – Jan 2019 | San Francisco Bay Area I worked with Pantheon, a website-hosting platform, through one of my organizations at Berkeley. We agreed to build a project manager dashboard for them through the semester using Go, React.js, and Firestore as our database. This dashboard combines several software management services including GitHub, Jira, and Pantheon's own platform data in order for project managers using Pantheon to more easily oversee progress and completed work. We received data from service webhooks on our backend, and simultaneously piped that data into our database and sent the data along websockets to our active clients for real-time updates on the dashboard. My role in this project was first as the project architect. Along with my co-project manager, we discussed and planned out the full tech stack, file organization, deployment/development pipelines, and any other smaller technical details. We onboarded 5 developers to the team and acted as project managers for the rest of the semester. We broke down the project into week-by-week tasks, provided technical workshops and office hours for our developers, and generally managed the project to ensure smooth and steady progress through the semester. ### Full Stack Developer @ BetterCloud Jan 2018 – Jan 2018 | San Francisco Bay Area I worked with BetterCloud, an operations management platform provider, through one of my organizations at Berkeley. I worked to integrate BetterCloud's service for creating/executing IT workflows into ServiceNow, another IT management tool - I used ServiceNow's extensive API to combine the two services into one seamless experience. For example, their GlideRecord API serves as a delegate to send and receive BetterCloud IT workflow objects to and from the database. This project follows a serverless architecture using Google Cloud Platform. ### Application and Systems Engineer Intern @ Prudential Financial Jan 2018 – Jan 2018 | Sunnyvale, CA At Prudential, I worked primarily as a backend developer on Prudential's new LINK application, a unified dashboard that brings together all of Prudential's financial services, built using the Java Spring Framework. In my time there, I worked on three major tasks: (1) I wrote new Java classes to validate/instantiate incoming data from the frontend into objects that were easy to work with in the backend. (2) I migrated project code that sends HTTP requests to updated service API endpoints, and wrote accompanying test suites for the migration. (3) I took a regularly executed, computationally expensive batch processing job that was responsible for updating over a million user profiles per execution and parallelized it using Apache Spark cluster computing. Performance benchmarks improved by over 8x, though a higher factor could easily be achieved if run on a larger cluster. ### Machine Learning Engineer @ Polymorph Jan 2018 – Jan 2018 | San Francisco Bay Area I worked with Polymorph, an ad-tech company, through one of my organizations at Berkeley. I developed on several machine learning models to answer the following research question: What is the probability a user visiting a site clicks a particular ad (also called Click Through Rate - CTR) based on demographic and other miscellaneous supplied information? Answering this question well would allow Polymorph to maximize their expected revenue. The model I worked on primarily employed a newer method particularly effective at calculating CTR called field-aware factorization. I also worked with models that employed logistic regression with SGD, random forests, and gradient boosting. In all, my team processed over 200GB of ad data with the field-aware factorization model achieving a log loss of around 0.48 or about 73% accuracy on a dataset with a balanced number of positive and negative samples. ### Computer Science Undergraduate Student Instructor @ University of California, Berkeley Jan 2017 – Jan 2018 | Berkeley, CA For three semesters, I taught Berkeley's well known and renowned introductory computer science class - CS 61A - that takes in over 2000 students per half. Each semester, I engaged with around 35 students through a lab and a discussion, as well as office hours and one-on-one advising. I partook in weekly meetings to assess and adjust the course as necessary to fit student needs and helped write exam questions for the midterms, review sessions, and finals. Fun fact: Once I guest lectured a summer session class of around 400 students on Scheme, a functional programming language and one of the three major languages taught in the class - see link! I start around the 21 minute mark. ## Education ### Bachelor’s Degree in Electrical Engineering and Computer Science University of California, Berkeley ### Whippany Park High School ## Contact & Social - LinkedIn: https://linkedin.com/in/matthewjeng - Portfolio: https://mjeng.github.io --- Source: https://flows.cv/matthewjeng JSON Resume: https://flows.cv/matthewjeng/resume.json Last updated: 2026-03-29