Extending a Silicon Valley Unicorn’s development capabilities

Overview

Industry

Location

Software and IT services

USA

Technology Used

Airflow

Bazel

Dropwizard

Elasticsearch

Hive

Java

K8s

Kotlin

Loki

Parquet  AWS

Postgres Spark

Presto

Prometheus Flink

Protobuf

Python

React.js FoundationDB

StackStorm

Tekton

Project overview

Software Mind is enhancing our client’s application development capabilities. We’ve been providing support in various areas, including infrastructure, tooling, backend development, DevOps, mobile development and data science.

Client background

Located in Silicon Valley, our client is a rapidly growing provider of cross-platform linking and attribution solutions. More than 100,000 companies have chosen to partner with our client to enhance their business, improve customer experiences and optimize campaign performance.

Developing Mobile SDK

Developing an Android SDK that enables advanced mobile search functionalities. The created SDK was embedded in OEM firmware and supports various search types, including application, file and content searches. Additionally, we incorporated smart ads and optimized their placement within the SDK, which is utilized by leading smartphone manufacturers such as Xiaomi, Vivo and Samsung

Data Engineering

Re-engineering, developing and maintaining the ad attribution data processing system. This system includes both batch and streaming data pipelines, along with various storage types. It efficiently processes approximately 10 terabytes of data daily without interruption. Additionally, it provides the data science team with easy and real-time access to data.

Data Science

Implementing a real-time ad auction system that updates over the air and operates on mobile devices with significant resource constraints. The designed system influences decision-making across both go-to-market and engineering teams by providing insights from collected data. Our work involved full-stack development, including on-device engineering logic and data-collection mechanisms, creating data processing pipelines, analyzing collected datasets, designing and running experiments, implementing changes and preparing business intelligence (BI) tools.

DevOps

Modernizing, optimizing and maintaining our clients' infrastructure. Software Mind’s task involved managing an AWS infrastructure comprised of over 1,000 Kubernetes (K8s) nodes. We also implemented a continuous deployment pipeline for a monorepo and introduced an efficient logging solution capable of handling hundreds of gigabytes of logs daily.

Results

20x growth in usage

SDK changes led to increased usage from 10 million devices to 200 million devices worldwide.

5x increase in revenue

Software Mind introduced various earning-impacting experiments, which led to a fivefold increase in revenue.

$1,000,000 USD saved

Executing cloud optimization that saved over $1,000,000 USD annually with no impact on system performance.

3x better detection

Three times as many bad records were detected and corrected due to the implementing dead-letter queues, ad hoc quality checks and statistical quality checks.