Brand Networks to Snap, Inc.¶
Snap Ads Manager¶
Our flagship product was a SaaS tool for marketing teams to manage ad campaigns. This included a power tool for users to upload and download detailed spreadsheets of tens of thousands of ads quickly to Facebook and Snapchat.
Snapchat purchased the software and hired us to integrate it with their platform in 2017.
Origins Analytics¶
Origins Analytics, Inc. developed relationships with gamefi studios PlanetIX, Matr1x, Mighty Bear Games, and partnered with Overworld Games. The platform and API combined on-chain and off-chain wallet statistics to boost player acquisition and retention for partners and customers.
From ingestion and pre-processing to segmentation and summary, organizing and optimizing for rapid updates and realtime consumption is a demanding process that relies on operational stability, speed of iteration, and meticulous QA. The pipeline ingested from several sources including proprietary game data, direct-to-node Ethereum and Polygon contract history, wallet portfolios and marketplace trade data.
Utilizing game-specific per-player rankings and activity, on-chain activity and marketplace trade data, users were segmented into concise cohorts.
Web3's multi-chain, multi-marketplace, bring-your-own API ecosystem experienced high-volume, fast-growth of nonuniform data. Creativity, drive, and efficiency in planning and execution were key to navigating challenges like standardization, sustainability, and prioritization. Web3 gameFi studios bring their own proprietary data and mix of web2 and web3 contexts to wrangle.
Thrive Market¶
- Introduction of AWS SQS and Lambda Function to the ecosystem (stage, prod, dev, local environments)
- Stop-the-world code issue with OOP solution
- Microservices migration and upgrade planning
- RCAs and permanent zero-regression fixes for several severe and long-standing issues
High frequency trading¶
- Custom stop limit orders
- Custom order book parser and realtime ETL
- West and east APIs over websocket to mitigate latency
- Negligible monthly cost
- Net profitable and $3k record week-high
ML Pipeline¶
- Lifetime value estimations
- Zynga, Fandango
Python scikit-learn Random Forest Classifier
Before: Jenkins, MySQL
After: Apache Airflow, PostgreSQL
- S3 sensors, source and sink
- Python dataclasses
- PostgreSQL
- Day-partitioned, auto-archiving
- Query optimization
- AWS, Terraform, Kubernetes, CI/CD