M.S. Computer Science
AUDREY DANG
SOFTWARE ENGINEER_
I build reliable full-stack systems, browser automation, and developer tools with measurable impact.
ABOUT
I'm a software engineer focused on building systems that are useful, observable, and dependable.
My work spans full-stack product engineering, backend infrastructure, browser automation, and ML-backed financial tools. I care about clean boundaries, thoughtful user feedback, and performance you can measure.
I'm currently pursuing an M.S. in Computer Science at UT Austin after completing my B.S. in Computer Science with a Mathematics minor and a 3.99 GPA at the University of Houston.
const audrey = {
focus: ["backend", "full-stack"],
values: ["clarity", "reliability"],
learning: "always"
};B.S. Computer Science · Math minor
EXPERIENCE
SOFTWARE ENGINEER INTERN
@ CodeDay Labs
- Built backend observability for Keyshade using Redis and NestJS, supporting 15K+ daily secret pulls and CLI executions while accelerating metrics processing by 30%.
- Fixed a production race condition in Maxun by serializing navigation flows and hardening async error handling, eliminating crashes in repeated concurrency tests.
- Standardized datetime and duration validation across Node.js APIs, reducing validation-related failures by 25%.
SELECTED PROJECTS
- 7 composable workflow node types
- Dependency-aware execution, retries, live updates, and cancellation
- One Browserbase session per run, reducing workflow latency by 37%
- 45% faster automation setup
OPEN SOURCE
Rapid recording-discard crash fix
Server-and-client crash path eliminated
Prevented overlapping page navigations when recordings are discarded in rapid succession, removing the ERR_ABORTED failure that crashed both server and client.
Real proxy-connection verification
Invalid DNS, connection, and credentials caught early
Changed the proxy test endpoint to launch a proxy-enabled browser context and report success only after a real proxied navigation succeeds.
Additional Maxun contribution (#1186)
Additional upstream contribution from the OSS record
The shared OSS document includes this pull request without a title or summary; the linked upstream discussion has the complete implementation and review context.
asset_key_prefix support for dbt sources
Reusable dbt source prefixes, with tests and docs
Let users define a shared asset key prefix in dbt_project.yml instead of repeating it across individual dbt sources.
Redis resource settings in the Helm schema
Configurable CPU and memory for internal Redis
Exposed the Bitnami Redis chart's existing resource settings through Dagster's Helm schema, making deployment sizing easier to configure.
Readable API usage metrics
15K+ daily operations · 30% faster processing
Added readable daily metrics for secret pulls, variable pulls, and other API activity, making production usage easier to monitor and act on.
Route checks for time and duration inputs
Centralized validation across relevant endpoints
Expanded route-level validation for timeInterval and duration inputs, centralizing shared checks before malformed values reach application logic.
Cross-platform screen uniqueness checks
Reliable screen tests on Linux Wayland
Made screen tests assert uniqueness by display origin instead of monitor name and skipped an origin assumption that Wayland does not guarantee.
Clarify server limits for client queries
Prevents incorrect assumptions about query limits
Documented that client query methods remain constrained by the server's default limit when limit=None, even when limit and offset are available.
SKILLS
Core stack
Backend
Cloud & infrastructure
Testing & tools
LET'S BUILD
SOMETHING USEFUL.
I'm open to 2027 software engineering opportunities and conversations about backend systems, developer tools, and open source.