Dragoneye helps devs build powerful apps and features that use images and videos. With our cutting-edge AI tech, folks can recognize things in the world with high accuracy and deep granularity, right out of the box. No more arduous process of annotating any training data or doing any machine learning work themselves.YCombinator W24

Dev on the Options desk

Tech Lead for the Places Search team. Worked to realign the team’s strategy into the company’s new focus on AR/VR use cases. Tech Lead for the Infra portion of the Local News team, working with 4 engineers focused on product infrastructure, content classification, and rankingWorked on our Local News surface as a 0 to 1 initiative, which grew to the largest local information provider in the United States. Worked on Local Alerts, a 0 to 1 product that helps local governments and first responders keep people in their communities safe.In 20% time, developed a statistical location understanding framework for ML that was adopted across the company to dramatically improve location relevance in products spanning News, Pages, Groups and Search teams. U.S. Patent 11,252,120 granted for framework.Led the implementation and roll out of a major revamp to our Local News product to a Feed experience. Major portions of the stack rebuilt on the latest recommendations frameworks at Facebook and coordinated the efforts across 14 engineers + XFN

Lecturer for CMSC 250: Discrete Structures - Summer 2017https://www.cs.umd.edu/class/summer2017/cmsc250/index.html

CIS 160 - Mathematical Foundations of Computer Science (Fall 2015, Spring 2016, Fall 2016)CIS 121 - Data Structures and Algorithm (Spring 2015) - TACIS 160 - Mathematical Foundations of Computer Science (Fall 2014) - TALed a course staff consisting of ~25 TAs, managed course logistics and created new class content to supplement existing material

• Worked with Professor Nadia Heninger on a fast and cost-efficient method of providing factorization as a service, with a focus on factorizing 512-bit keys as used in protocol downgrade attacks such as FREAK and Logjam• Research paper (http://eprint.iacr.org/2015/1000) accepted to Financial Cryptography and Data Security 2016 as a full paper, and featured in Ars Technica (http://arstechnica.com/security/2015/10/breaking-512-bit-rsa-with-amazon-ec2-is-a-cinch-so-why-all-the-weak-keys/). Code published at https://github.com/eniac/faas• Explored the performance, reliability and scope of number field sieve (NFS) libraries CADO-NFS and Msieve• Gauged the feasibility and limitations of using OpenMPI for HPC on commodity clusters on Amazon EC2• Developed factorization scripts in Python in response to testing results, integrating the most efficient parts of each library into a complete factorization package• Parallelized the existing C code for the square root phase of the Msieve library, moving from a non-deterministic runtime to a deterministic runtime that was two to four times faster on average• Tested systematic modifications to NFS factorization parameters and instance/cluster configurations, allowing for the optimization of cost, factorization speed, and reliability tradeoffs

Worked on the Account Access team. Work included launching a new two-factor authentication method on Android, speed up of logout time on Facebook for Android, and redesign of the two-factor authentication checkpoint for login.