David Yu
CS at Northeastern, class of 2027. Lives in New York City.
GitHub LinkedIn davyu735@gmail.com
Experience
- Software Engineer Intern @ LinkedIn, May 2026 - Aug 2026
- Automation Software Developer Co-op @ Citizens Financial Group, Jul 2025 - Dec 2025
- Software Engineer @ Generate Product Development Studio, Jan 2025 - Apr 2025
- Software Engineer Intern @ Amazon, May 2024 - Aug 2024
Projects
- IntReview: An AI-powered interview preparation platform that provides real-time feedback on verbal and non-verbal communication using computer vision and speech analysis.
- Real-Time AI Business Intelligence: Ask your business data questions in plain English and get SQL and charts back. An MVP that is still being built.
- Constellation: An iOS app that visualizes your social network as an interactive galaxy, using on-device machine learning to predict which relationships need attention and encouraging meaningful reconnection.
- Black-Scholes Options Pricing Model: A sophisticated financial tool implementing the Black-Scholes mathematical model to accurately price options contracts and analyze market volatility.
- Classify: A machine learning application that intelligently categorizes Spotify playlists into themed collections to enhance music discovery through audio feature analysis.
- PlateMate: A mobile application that revolutionizes restaurant reviews by focusing on what truly matters to diners - the quality of individual dishes.
- Distyl: A VS Code extension that figures out which parts of your workspace an AI coding assistant actually needs, then packs them to fit inside the context window.
- Multi-Agent Personal Automation System: Four agents running around the clock on my own machine: one runs the day, one handles code, one researches and publishes content end to end, one handles career search and outreach. No database, no server, and no memory of anything except what they write down.
I'm a CS student at Northeastern who gravitates toward problems where the interesting part isn't the code. Most of what I've built started as a small annoyance: interview prep that gives you no real feedback, restaurant reviews that rate the room instead of the food, an inbox that buries the five emails that matter under three hundred that don't.
I've spent my last few summers at Amazon and LinkedIn learning how software behaves when a lot of people depend on it, and my semesters building things where I get to make every call myself. Those two teach very different lessons and I'm not finished with either. Lately I'm drawn to early-stage teams, where the distance between deciding something and shipping it is measured in days.
Software Engineer Intern
Built
- A new Profile feature spanning 6 codebases and 3 native clients (web, iOS, Android), driven entirely by one server-defined UI description.
- Designed the scoring API as a stateless, deterministic service, so replay safety comes from the inputs rather than session tokens.
- Modeled persistence as a three-state lifecycle with conditional writes, backed by storage-layer invariant tests.
- Turned the privacy and authorization requirements into build-gating tests instead of review checklist items.
Outcomes
- Collapsed three parallel client implementations into one source of truth that ships without a client release.
- Removed every server round-trip until final submission, so the feature stays responsive with no reconciliation bugs.
- Guaranteed by test that retakes commit zero writes and duplicate requests land on identical end states.
Tech
JavaKotlinServer-Driven UIREST APIsiOSAndroidIntegration Testing
Citizens Financial Group
Automation Software Developer Co-op
Built
- A UiPath package management tool with dev-lead approval workflows for deploying enterprise automations.
- An automated Nexus deployment pipeline distributing 50+ development packages.
- Azure VDI scripts for virtual desktop operations.
- Part of a cross-functional GenAI initiative and responsible-AI governance work.
Outcomes
- Cut deployment time by 60% with streamlined approval workflows.
- Cut manual processing time by 25% across 100+ virtual desktop environments.
Tech
UiPathAzure VDINexusCI/CDEnterprise AutomationGenAI
Generate Product Development Studio
Software Engineer
Built PlateMate, restaurant reviews for the dish, not the room.
Built
- REST API in Go/Fiber on MongoDB, tuned to sub-100ms responses.
- Auth with JWT and role-based access, so verified owners manage only their own menus.
- Designed the client-server split so the React Native app and Go backend could evolve independently.
Outcomes
- Shipped PlateMate to TestFlight in 3 months, covering 100+ Boston restaurants.
- Owned it end to end, from Figma prototype to backend, testing and deployment.
Tech
GoFiberMongoDBRESTJWTReact Native
Amazon
Software Engineer Intern
Built
- Extended a production AWS Lambda from one supported system to three, automating health reporting for two experimentation and content-optimization platforms that were reported by hand.
- 7 priority-zero metrics: workflow success rate, execution volume, P50/P90/P99 batch runtime.
- Refactored a monolithic Lambda into shared, unit-testable functions so all 7 metrics run through one path.
Outcomes
- Took two platforms from zero automated health coverage to recurring weekly and monthly reporting.
- Derived execution volume for a system with no direct telemetry by modeling hourly workflow runs across 15 marketplaces at 360 runs/day.
- Merged after 7 rounds of review, output checked against manual baselines.
- Presented on cloud cost optimization to the GCCP team.
Tech
JavaAWS LambdaCloudWatchS3
scroll to open the laptop: where I've worked⌄
flat
zig-zag over the bowl to whiskswipe back and forth on the bowl
currently listening to
Back To U · SLANDER, William Black ↗