Brainrex joins Libracamp, a boot camp to build on Facebook’s backed cryptocurrency project.
After 4 weeks of voting, Brainrex has been selected as one of the 5 teams out of 160 that will be participating in Libracamp, a 2-month online boot camp to accelerate companies building on top of the Libra blockchain testnet.
Libracamp is an online accelerator that helps startups building on Libra. The upcoming Libracamp platform will connect Libra developers with mentors, grant opportunities and investors, offering a stronger “crowd-sourced” vibe than most accelerator programs.
Anomaly detection. We automatically find errors in your data by applying machine learning.
Natural Language Processing. Offers sentiment analysis, entity analysis and categorization technologies, capable of understanding the general feeling/attitude of in the written text.
We built integrations to everything.
-Cloud warehouses: AWS S3, Redshift and Gcloud BigQuery, Cloud Storage. Kinesis.
-Blockchain Data: Ethereum, Ethereum Classic, Bitcoin, Bitcoin Cash, Monero, Zcash, Litecoin, DOGE, Libra.
-Cryptocurrency Data: 23 exchanges and 1500 currency pairs currently supported through third parties e.g Shrimpy API, Kaiko.
-Social Media Streams: Twitter, Reddit, Google Trends and RSS channels.
A set anomaly detection pre-trained models that are embeddable in the Libra blockchain network and applications. The libra sentiment score. We will collect and clean as much news articles, posts and mentions about Libra, and perform sentiment analysis on it.
Sprint 1: Collecting data. (Nov 4th — Nov 15th) We will build a data parser that connects to GRPC feed and uses Apache Kafka to make some transformations and saves it in a scalable data warehouse for late use in training and testing phase. A document store database will be used to . Server space will be provided for free by our friends at IBM.
Sprint 2: Simulating post-launch usage in the Libra testnet(Nov 18th — Nov 29th ). We will build three kinds of bots that will either mint coins, transact coins, launder money. These bots will be joined by equivalents malicious bots that will attack the testnet.
Sprint 3 (Dec 2nd — Dec 20th): Training neural network for anomaly detection. We will use the data collected in phase one and two to train. Feature extraction will take most of the time. A deep neural net architecture will be designed and will be trained in Tesla v100 GPUs. The model will be then made available through the Brainrex API and as an embeddable file capable of running in the Move programming language.
Let’s get Libra off the ground!