Staff Software Engineer, Feature Platform

Unity Unity · Enterprise · Mountain View, CA · Templates

Staff Software Engineer on the Feature Platform team at Unity, responsible for building and operating infrastructure that powers machine learning, experimentation, and optimization for their ads ecosystem. This involves designing and operating systems that transform high-volume event data into production-grade feature datasets for bidding, attribution, and ranking, working at the intersection of distributed systems, platform engineering, and ML infrastructure. The role owns the full software lifecycle of pipeline systems, supporting both offline training and online feature serving.

What you'd actually do

  1. Design, build, and operate scalable, production-grade data pipeline systems and curated feature datasets powering ads optimization and ML
  2. Own end-to-end offline data flows from raw event ingestion to feature-ready datasets, with strong emphasis on correctness, reproducibility, and SLA compliance
  3. Develop and maintain large-scale batch and streaming systems (Python / Java / SQL) with a strong focus on performance, cost-efficiency, and reliability
  4. Build and contribute to our Feature Store platform, including integration with the high-throughput online serving layer (Go-based services)
  5. Translate complex product and monetization logic into well-engineered, extensible systems serving analytics and machine learning use cases

Skills

Required

  • Python
  • Java
  • SQL
  • Spark
  • Flink
  • distributed systems
  • data pipelines
  • ETL/ELT
  • cloud-native environments
  • containerized systems
  • Kubernetes
  • workflow orchestration tools

Nice to have

  • Go
  • ML infrastructure
  • feature stores
  • model training pipelines
  • ads
  • attribution
  • monetization systems
  • experimentation and metrics infrastructure
  • high-scale backend or platform engineering

What the JD emphasized

  • Strong software engineering fundamentals with deep experience designing and operating large-scale distributed systems in production
  • Hands-on experience building production-grade ETL/ELT pipelines using Python, Java, SQL, or similar technologies
  • Experience with distributed processing frameworks such as Spark or Flink in both batch and streaming modes, including performance tuning and parallel computation
  • Understanding of how offline data systems integrate with online serving layers — feature stores, low-latency APIs, and real-time systems
  • Strong ownership mindset — focused on correctness, observability, and long-term maintainability

Other signals

  • building infrastructure that powers machine learning
  • transform high-volume event data into production-grade feature datasets
  • offline training workflows and online feature serving
  • feature store platform
  • integration with the high-throughput online serving layer