Welcome to my homepage!
Hello, my name is Horatiu Lazu, currently studying Computer Science at the University of Waterloo. I'm currently a Software Engineering Intern at Google. Formerly at Facebook in Seattle, Citadel Securities in Chicago, Yelp in San Francisco, twice at Bloomberg LP in New York.
Road cycling is my hobby, with my Strava Wind Analysis site providing analytics and data visualization to over 120,000 users.
Feel free to visit my Blog, GitHub, or my YouTube Channel that has over 1.4 million views.
Road cycling is my hobby, with my Strava Wind Analysis site providing analytics and data visualization to over 120,000 users.
Feel free to visit my Blog, GitHub, or my YouTube Channel that has over 1.4 million views.
In Winter 2020, I joined Facebook in Seattle, Washington on the Messenger Real-Time Signaling team. The Messenger RT Signaling team is responsible for all real-time communication signaling used by Facebook's family of products excluding WhatsApp (which includes Messenger and Instagram).
My primary project was redesigning the synchronization and outbound APIs for the multiway calling service. Using a pipelined architecture, the different paths used to establish and maintain hundreds of millions of calls daily now allows product teams to have cleaner/more efficient control flow and instrumentation.
In Summer 2020 I joined Google for a remote internship (originally based out of Mountain View, California) on their Firebase SDKs team. Firebase is Google's mobile platform that helps you develop high-quality apps, known mostly for their real-time database but now consists of 18 other products as well.
During my first month I built and shipped the ability to bulk import users into Firebase in C#. For the remainder of the internship, I am focusing on large-scale refactoring to modularize and auto-generate typings from the source for the Node.js Admin SDK.
During my first month I built and shipped the ability to bulk import users into Firebase in C#. For the remainder of the internship, I am focusing on large-scale refactoring to modularize and auto-generate typings from the source for the Node.js Admin SDK.
In Summer 2019 I joined Citadel Securities in Chicago, Illinois to work on the trading platform in C++. By combining low-latency software and quantitative research, Citadel Securities trades over a billion stocks daily constituting ~40% of US retail equities trade volume and is the largest market maker in the US.
Learn more about HFT (High-Frequency Trading) and Citadel.
Learn more about HFT (High-Frequency Trading) and Citadel.
In Fall 2018, I joined Yelp on their Production Engineering Team in San Francisco, California. The Production Engineering Team is responsible for handling the scale of over 100 million monthly active users on Yelp, which includes lower level systems and infrastructure.
I developed and shipped a scalable cloud metadata search service using Elasticsearch, Python, Terraform and AWS. This service allows for expressive discovery using a custom defined DSL, features a RESTful interface powered by a serverless architecture and provides an elegant dashboard written in React.
I developed and shipped a scalable cloud metadata search service using Elasticsearch, Python, Terraform and AWS. This service allows for expressive discovery using a custom defined DSL, features a RESTful interface powered by a serverless architecture and provides an elegant dashboard written in React.
In Winter 2018, I returned to Bloomberg LP for another term in New York! I joined the Fixed Income Core Search Team, which is responsible for powering all fixed income search related infrastructure and functionality across Bloomberg products.
I worked on contributing the ability to perform high-performance distributed joins across sharded collections and SolrCloud clusters via ZooKeeper host to Apache Solr. Using various optimizations including leveraging uninverting Lucene data-structures, extensive performance profiling,use of efficient data-structures (roaring bitmaps) and compression techniques: the average join query time was reduced by over 88.7% compared to streaming intersect expressions.
Read more about this experience in my blog post.
I worked on contributing the ability to perform high-performance distributed joins across sharded collections and SolrCloud clusters via ZooKeeper host to Apache Solr. Using various optimizations including leveraging uninverting Lucene data-structures, extensive performance profiling,use of efficient data-structures (roaring bitmaps) and compression techniques: the average join query time was reduced by over 88.7% compared to streaming intersect expressions.
Read more about this experience in my blog post.
In Summer 2017, for my first co-op term, I went to Bloomberg LP in New York. I joined the Excel Add-In team, responsible for bringing real-time and historical financial data from the Terminal straight into their Excel spreadsheets and perform financial analysis.
My first project was developing a machine learning collaborative filtering recommender system to generate personalized recommendations for the 250,000+ monthly Excel Add-In users in terms of which Excel Add-In functions they should explore. For my second project I worked on prototyping a chatbot to compose Bloomberg Query Language queries for equity and bond questions posed by users and fetching the respective results.
Read more about this experience in my blog post.
My first project was developing a machine learning collaborative filtering recommender system to generate personalized recommendations for the 250,000+ monthly Excel Add-In users in terms of which Excel Add-In functions they should explore. For my second project I worked on prototyping a chatbot to compose Bloomberg Query Language queries for equity and bond questions posed by users and fetching the respective results.
Read more about this experience in my blog post.
In Summer 2016, I joined Bluesun in Appleby Ontario for an internship. Bluesun is the largest provider of insurance software (WealthServ) in Canada for the private channel. I am very grateful for the opportunity I had at Bluesun, since I had only graduated high-school but they were still willing to take me on board.
My project for that summer was creating a multi-threaded wrapper of the git-tf migration tool in C# to convert all their TFS (Team Foundation Server) repositories to Git. The wrapper allowed for queuing, fixing working tree file corruptions, and executing branches as seperate threads improving performance by up to 70%.
Strava Wind Analysis
Provides historical and aggregate ride/segment analytics, data visualization and machine learning for cyclists on the Strava platform
Karnaugh Simplification
Simplifies boolean expressions visually.
Slope Field Generator
Graphs slope fields from seperable differentials.
Mandlebrot Set Generator
Generates fractals.
Synthetic Division
Demonstrates Ruffini's extended law of division.
Atomic Smasher
Demonstrates the basics of chemistry in a game.
Segment Analysis
Calculates wind influence on cycling rides.
Basilisk 3D
Snake game with various game-modes.