Portfolio

Brilliant.org

Full stack developer
Over 1 million registered users
Social+Capital backed startup
5th Engineering hire

Brilliant is a massive community of math and science problem solvers.

Key Achievements:

Led a charge that decreased average page load times by 40%.
Implemented Require.js on a 3 year old codebase that relied heavily on global scope.
Helped the product grow from under 200k users to over 1 million.
Doubled the amount of items shared on social media and created sharing framework.

Technologies Used:

Python
Django
Javascript
JQuery
Require.js
LESS
Git

Features I worked on:

User submitted problems
Newsfeed
Following
Resharing
User submitted solutions and comments
Wiki articles and practice problems

Online Calculator
Partial page loading framework
User profiles and moderation
Sharing on social media
Latex processing
Building and deploying

401(k) Investment Due Diligence

While employed by XRM.com (Streamline Solutions) and working for Retirement Plan Advisory Group a division of 401(k) Advisors Inc. I built the industry leading Investment Due Diligence and reporting web application for Enterprise 401(k) Financial Advisors.

"Streamline Solutions has never let us down when it comes to providing exceptional work that is always affordable accurate and efficient" - Case Study.

Here's some of what it can accomplish:

Conduct In-depth Investment Analysis
Automatically Rank Funds with Proprietary Scorecard System
Built-in Investment Logic to Assist in Decision making
Produce Integrated Investment Policy Statement
Leverage a CFA-Led Research Team
Select 20,000+ Investments From 36 Asset Classes and 50+ Vendor Products, including Open Architecture
Monitor Asset Allocation Funds and Models
Integrate Market Summaries and Research into your Customizable Report
Access Secure Online Cloud Storage, The Fiduciary Briefcase, which Allows Committee Members to Access Investment Reports 24/7.
Investment Due Diligence can be Applied to 401(k), 403(b), 457, Non-Qualified and Defined Benefit Plans Ranging from $1,000,000 to $1,000,000,000+

Contact 401(k) Advisors to Sign up.

Investment Reviews and Scorecards are 100% customizable through the web application.

Sample Investment Review including Scorecard and Fund Fact Sheets

My Responsibilities

Lead Developer using C#, Javascript, SQL
Communicate with Customer
Manage Team of 4-7
Prioritize Tasks
Estimate Time and Cost
Maintain Site Reliability
Design Software and Hardware Architecture
Code Reviews and Performance Evaluations

Technologies used

C#
ASP.NET
MSSQL
Javascript
JQuery

Micorsoft Server/IIS
PHP
MySQL
iTextSharp
MPI Stylus

Noteable Achievements

Saved client $300,000 by meeting their deadline to convert fund importer from one data provider to another
Converted web application from PHP to C#/ASP.NET
Integrated Markov Process International Stylus system
Managed XRM's oldest and largest customer for over a year with no oversight
Implemented many new features in parallel while ensuring site uptime and on time data availability with only a small team
Created tools to automatically integrate many different types of funds from many different sources in different formats into a single unified database each quarter

Interstate: SIP Fuzzer

As a researcher at UC Irvine I built the initial version of a multi-threaded state-full fuzzer used to find vulnerabilities in SIP phones.

A resulting paper was published at Defcon 15 by Dr. Ian Harris.

"Fuzzing" is the act of inserting incorrect data into a message in the hopes of producing an error in the software receiving the message

"SIP" stands for Session Initiation Protocol and is an open source protocol widely used in Voice Over IP (VOIP) phones

Fuzzing SIP is very important because it allows researchers to find and fix vulnerabilities in the protocol and its implementation that could otherwise be exploited by nefarious parties trying to listen in to or otherwise manipulate a secure phone call.

Features:

Written in Python
Multi-Threaded (The system can both send and receive information simultaneously)
State-full (System has knowledge of which state the phone is in and can fuzz different types of messages accordingly)

The following papers describe the fuzzer, the SIP protocol and the results of the team's tests including the vulnerability found

INTERSTATE: A Stateful Protocol Fuzzer for SIP

Defcon Slides

Paramount Pictures Transaction Approval Process

While employed by XRM.com (Streamline Solutions) and working for Paramount Pictures, I built a transaction approval and inventory management system which tightly integrated with Microsoft Dynamics CRM.

“Streamline Solutions went above and beyond the basic needs of our project!” - Case Study.

Global approval system for purchases and return orders
Secure and reliable auditing through automated email processing
Tight integration with Microsoft CRM through Microsoft Outlook and a custom web application.
API integration with back end SAP system
Multi-threaded architecture provided 10x improvement over old system
Worked directly with the client and led a team of 3 developers

Music Artificial Intelligence

Music creation software which uses artificial intelligence to compose music.

Features:

Multiple cooperative independent agents
Music composed in MIDI standards for portability and interesting I/O possibilities
The algorithm attempts to emulate how Jazz musicians improvise or how birds flock in that no one agent has complete control yet collectively they determine how the music progresses

History: This project initially began as a semester long project for my Advanced Projects in Artificial Intelligence course at UC Irvine.

Future: Current work is being done on connecting the software to real world I/O such as synthesizers and drum machines. Initially to allow for better sounding sounds then the current software synths, but long term the goal is to allow user input to communicate and play in real time with the AI agents. Additionally, more advanced music theory, genres and visualizations are also planned.

Technology:

Multi Threaded Java
Java JSound
MIDI

The source has been open sourced and contributions are welcome on Github

Multi Agent Music AI | View the Code

Let's Eat

LetsEat.io is a real time collaborative single page web application for deciding on a place to eat amongst a group of people.

Here's how it works

Users create a room with the name of their company or group
Next the group is given a URL which they can share with their friends or colleagues
Users can search Yelp for whatever kind of food or restaurant they are interested in
Nearby results are displayed on a map
Users can chat in real time and vote on where they want to go
An adjustable clock counts down the time until its time to eat, at which point users are emailed the decision

What's for lunch? | letseat.io

Technologies Used:

Meteor Client and Server side Javascript Framework
MongoDB NoSQL Database
Handelbars Templating
Yelp API
Leaflet for Mapping
Git

Built by myself and launched to the public in early 2014

Jeff Kile

Brilliant.org

401(k) Investment Due Diligence

Interstate: SIP Fuzzer

Paramount Pictures Transaction Approval Process

Music Artificial Intelligence

Let's Eat

Bouncer

Creative Programming for Digital Media & Mobile Apps

C++ Art