Staff Engineer

The candidate will have responsibilities across the following functions:

Own the architecture of the money path:

• Design a demand-source-agnostic serving layer: any exchange, ad server or deal plugs in behind one contract we own (OpenRTB 2.6 with Glance extensions).
• Keep decisioning — auction, floors, priority and ad-load policy — in our platform, never in adapters or partner consoles.
• Define the rendering contract the client SDK depends on (native, VAST, MRAID/HTML).
• Make the big calls and write them down: ADRs, latency budgets, build-versus-use decisions with InMobi platform teams.

Serve ads inside the Glance experience:

• Build ad slots into Glance's content feed, with ad load balanced against dismiss rate, session depth and retention.
• Design for prefetch: ads fetched ahead of time and shown later, sometimes offline - caching, expiry, pacing and delayed impression beacons all have to hold.
• Own per-user frequency and ad load across every demand source, within latency, battery and memory budgets on low-end Android.

Set the bar and grow the team:

• Run design reviews across serving, SDK, automation and revenue data, and own the API contracts between them.
• Make "what proves this? " the first question on any pricing or serving change, AI-assisted code included: replay evals, invariant checks, auto-rollback canaries.
• Be accountable for integrity gates: creative safety, invalid traffic, policy compliance.
• Review revenue-event and partner-reconciliation design so the numbers hold up to OEM audit.
• Mentor senior engineers into area leads, and make sure every service has a second owner

Requirements:

• 10+ years, including ownership of a low-latency, high-throughput system where a bad deploy shows up as lost revenue - ad serving, payments, trading or pricing.
• Distributed systems depth: partitioning, replication, consistency, failure modes, graceful degradation.
• Deep, hands-on Java and Spring Boot (our serving stack), including JVM/GC tuning for low latency; comfortable in Python (FastAPI) for ML and data services.
• Cloud (GCP or AWS), Kubernetes and CI/CD, designing for cost of serving, not just uptime.
• Event pipelines at scale (Kafka, Spark or Flink), reconciled against an external source of truth.
• Safe releases without a manual pass: contract and replay testing, canaries, automated rollback, SL Os; you've been paged on RFCs and ADRs that lead to decisions, and influence across teams you don't manage.

Nice to have

• Sell-side ad tech: SSPs, ad servers, Prebid / Prebid Server, Google Ad Manager.
• OpenRTB, VAST, MRAID and OM SDK; privacy signals (T CF, GPP) and supply-chain transparency (schain, sellers.json, app-ads. txt).
• Android or mobile SDK work under size, battery and cold-start budgets.
• Consent frameworks (DPDP, GDPR, LGPD), invalid-traffic defences, or partner-billing systems that external partners audit.