Alex Dean
At AWS, I work on supporting and extending functionality for EBS snapshots. On the
team, I'm the service owner for EBS snapshot functionality in
Outposts
and Local Zones,
offering EBS snapshot support for large organizations, governments, and other customers
at edge data centers. These capabilities enable customers to run special workloads
for data residency and isolation, local data processing, and low-latency compute purposes.
In addition, my team and I work on the following core functions:
-
Metering — Sifting through trillions of snapshot
chunks worldwide on a daily basis to accurately bill customers for their
hourly snapshot usage
-
Garbage Collection — Identifying billions of
unreferenced snapshot chunks daily and marking them for deletion,
freeing up space for Amazon S3
-
Durability — Offering eleven 9s of durability for
exabytes of snapshot data in S3 by performing data integrity checks and
recovering chunks in failure cases
-
Recycle Bin — Providing accidental deletion protection
for EBS snapshots and EC2 Amazon Machine Images (AMIs)
Personal Projects
This list is ordered from most to least recent.
-
Seattle Pets — Web visualization that uses data from
the Seattle Open Data portal to showcase which areas of Seattle have the most
pets. The site was built using React for visualization and OpenStreetMap
for maps support. The original data CSV was processed and filtered down
to a JSON to speed up data loading and visualization. A GeoJSON is used to
center the circles along Seattle ZIP codes to indicate how many pets the area
has. You can see the website
and the GitHub page.
-
Satellite Sonnet — Application that generates a daily
Shakespearean sonnet about NASA's astronomy image of the day using AWS Lambda.
The application uses NASA's public API to get the day's image, encodes the
image into byte blocks, and sends the information to Anthropic's Claude for
sonnet generation. The result is cached into DynamoDB to minimize further API
calls. Upon users visiting the site, an atomic counter is incremented (using
AttributeAction.ADD) to inform users of what "visitor number" they are. You
can see the website, the
frontend code, and
the backend code.
-
Write and Compare — Anki add-on that provides the Write
and Compare note-type, allowing users to write multi-line answers to Anki cards
and test their knowledge in a more active format. Instead of passively reviewing
a card, this note type forces users to write down their answer for the question the
card is posing, resulting in slower and stickier learning. You can see the
GitHub page and
AnkiWeb add-on.
-
Generative Documentation & Testing — NPM package that
functions as a CLI tool using the GPT 3.5 API to improve your JavaScript and
TypeScript code. It helps you write better code by automatically generating
documentation, refactoring code to improve performance and maintainability,
adding unit tests using Jest, and converting JavaScript code to TypeScript.
At its peak, the package had about 700 weekly downloads. You can see the
NPM package
and GitHub page.
-
Generative Satire — Automated satirical news site monitoring
recent events and generating Onion-style articles using LLMs; At its peak,
the site had a viewership of about 1500 people per week. The application was
built using a React frontend and a Node.js backend, with articles stored on a
Firebase Realtime Database and multiple APIs used for image and content
generation. The site is hosted on Google Cloud. You can see the
website and the
GitHub page.
-
Read This to Me — Application that converts written text into
audio from a wide variety of dialects, using the SpeechSynthesis Web API. The
site was built using JavaScript and HTML/CSS. The site is hosted on GitHub
Pages with a custom domain. You can see the
website and
GitHub page.
-
Language Discovery — Application that applies various
similarity algorithms on various human language data sets. By languages,
I'm referencing frequency dictionaries for human languages, which catalog
how frequently a word is used in one language relative to another. I examine
topics like the average word length and the most popular color in a given
langauge. You can see the
website
and GitHub page.