> Markdown version of [/jobs/ext/4616635-senior-software-engineer-video-performance-bay-area-only](https://www.wearedevelopers.com/jobs/ext/4616635-senior-software-engineer-video-performance-bay-area-only). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Senior Software Engineer - Video Performance - (Bay area only)

- **Company:** CANVA L.C.
- **Location:** San Francisco, CA, United States
- **Experience:** Expert
- **Salary:** $262,000 - $329,000
- **Contract:** Permanent contract
- **Skills:** Application Programming Interfaces (APIs), Android Studio, C++ (Programming Language), Profiling, FFmpeg, OpenGL, Open Source Technology, Backend, Perf (Linux), HLSL, Hardware Acceleration, TensorRT
- **Published:** November 7, 2026
- **Apply:** https://www.juju.com/job/28_25196_01607c3022cb4781e4e3ebe08171baea

## About the Role

* Hands-on iOS or Android work focused on system-level performance: responsiveness, startup time, memory, battery and thermal behaviour. Experience with older Android devices is especially useful to us.
 * A working understanding of video pipelines: codecs, container formats, and frame-accurate seeking and composition.
 * On-device ML performance work, such as tuning inference latency with CoreML, TFLite or TensorRT.
 * Contributions to open-source multimedia or graphics projects such as FFmpeg or Chromium.

## Description

As Canva scales, change continues to be part of our DNA. But we like to think that's all part of the fun. So this will give you a flavour of the type of things you'll be working on when you start, but this will likely evolve.

 * Own performance across the whole video creation lifecycle: template load, media import, timeline scrubbing, real-time preview, effects and export, on web, iOS and Android. When a bottleneck turns out to live in someone else's layer, you follow it there.
 * Define the performance metrics and service level objectives for Video, and make sure they track what people actually notice rather than what happens to be easy to measure.
 * Build the telemetry, tracing, dashboards and alerting that keep video performance visible, including regression detection that catches problems before users feel them.
 * Build the profiling and benchmarking tooling: traces, device labs, automated performance suites. The goal is that any engineer in Video can diagnose a slow path without waiting for you.
 * Ship measurable improvements to latency, frame stability, memory and battery life. Sometimes that means a product or pipeline change. Sometimes it means critical-path work in the native engine, the shaders, or the ML-powered effect pipelines.
 * Work with the Native Video Engine, on-device ML and Editing teams to rework hot paths for better parallelism, caching, hardware acceleration and asynchronous processing., * C++, or another systems language such as Rust, used to optimise critical paths in a native engine. Plus a feel for how graphics APIs (OpenGL, Metal, Vulkan, WebGPU) affect frame times.
 * You've made real systems measurably faster, and you can talk through the numbers.
 * You move comfortably between product code, backend services and native or platform layers, and you can find your bearings in an unfamiliar stack.
 * You've built performance tooling that other engineers came to rely on: profilers, tracing and instrumentation, benchmarking harnesses, regression suites or telemetry pipelines. You're at home in tools like Perf, Instruments, Perfetto, Android Studio Profiler and Chrome DevTools, and happy to build your own when none of them fit.
 * You can define a metric that means something, read telemetry at scale, spot a regression, and prove an optimisation worked.
 * You understand concurrency, memory behaviour and CPU and GPU cost models well enough to explain why something is slow, not only where.
 * You prioritise by user impact, starting with the journeys people use most on the devices that struggle most, and you can get performance work landed by teams you don't manage.