Staff Software Engineer, Instawork Robotics

Instawork runs a marketplace for hourly shift work, and Instawork Robotics Labs is its newer bet: sending skilled workers into real kitchens, warehouses and hotels to capture the physical task data that robotics labs use to train foundation models. This role owns the engineering behind that at the Bengaluru hub, building Instawork Lens, the global data collection platform, and Prism, the QA platform. It is a genuine generalist seat. You move between frontend, backend and infrastructure, architect AWS and ECS environments for video transcoding, and design pipelines that move petabytes of video and telemetry without falling over. The posting sets an explicit expectation that you are still coding around 80 percent of the time, which is worth noting for a role at this level.

What the posting requires: 8 or more years of software engineering experience. Deep understanding of distributed systems, database internals and network constraints, and clean maintainable code, with the posting stating you will be coding roughly 80 percent of the time. Extensive full stack experience with Python, Django, React or similar modern frameworks, and willingness to pick up whatever language the job needs. Hands on cloud and infrastructure experience: AWS, ECS and Docker, CloudWatch, particularly for data heavy or high throughput applications. A startup mindset, described as thriving in ambiguous zero to one environments with high velocity decision making.

The day to day: owning features end to end across the stack; leading technical development of the Lens and Radiant platforms for data enrichment, labelling and continuous QA; architecting and scaling cloud environments and ECS services for video transcoding with telemetry across CPU, memory, network and disk IOPS; designing petabyte scale data pipelines; using agentic models as a force multiplier while making the architectural calls yourself; and mentoring mid and senior engineers through code review, pairing and architecture discussion.

Location and office reality: Bengaluru, described as a hub the company is building out.

Who this is for: a true generalist who is comfortable being the technical pillar of a small new team and does not want to specialise. The physical AI and robotics data angle is genuinely different from most listings here, so it suits someone who wants that domain. Backed by Benchmark, Spark Capital, Craft Ventures, Greylock and Y Combinator, if investor quality matters to your risk calculation.