What You'll DoBuild the orchestration layer. Design and implement a scalable workflow system to ingest, route, and process incoming events. Define the stages of the pipeline - ingestion, preprocessing, inference, validation, and delivery - and build something that handles failures gracefully at high throughput.
Scale the inference fleet. Build the compute layer that parallelizes processing across the event backlog and handles burst capacity as our camera fleet grows. Design the worker pool, queueing, and autoscaling strategy for GPU-bound workloads on ECS.
Design the data plumbing. Own the path from edge device to pipeline output - storage, metadata, and the triggers that drive processing. Build something that is observable, debuggable, and auditable end-to-end.
Build the model serving and lifecycle layer. Stand up the infrastructure that loads versioned CV models and handles inference reliably. Optimize for GPU utilization and throughput where it matters - dynamic batching, multi-model serving, and model optimizations like quantization or TensorRT/ONNX. Ensure new model versions can be promoted and rolled back without pipeline downtime.
Set the engineering standard. This is an early hire. You'll write the playbooks - runbooks, deployment procedures, testing standards - that the team builds on as we grow.
What We're Looking ForDepth in backend systems. 6+ years building and operating production backend or data-intensive systems at scale, with meaningful experience working on ML-heavy pipelines. You've owned something through its full lifecycle - design, deployment, scaling, and on-call - and you've done it in a context where ML inference was a first-class part of the system.
Hands-on orchestration experience. You've used a workflow orchestration tool to build production pipelines, not just evaluate them. You understand the tradeoffs between options and can make a principled choice for our use case.
Strong cloud infrastructure fundamentals. Comfortable with the building blocks - compute, queues, storage, networking - and you think in terms of cost, reliability, and operational simplicity rather than just what works.
Enough ML systems fluency to orchestrate them well. You've built or operated pipelines where ML inference is a core stage, and you understand what those workloads need - throughput constraints, GPU economics, model versioning, and keeping model performance visible in production. You don't need to have trained the models, but you know how to run them reliably at scale. Experience with CV or video pipelines is a plus.
Pragmatic decision-maker. You don't reach for the first framework you know. You understand the problem, evaluate tradeoffs honestly, and build something that fits the actual scale and constraints.
Why ObvioMission. Your work will directly help save lives and improve road safety.
Ownership. You'll make foundational architectural decisions during a critical phase of our Series A growth.
Impact. This isn't a maintenance role. You are being hired to build the core ML infrastructure layer from the ground up.
Growth. Competitive compensation, early-stage equity, and the opportunity to build a world-class ML platform organization.
Why Obvio- Your work will help save lives and improve road safety
- Series A of $22M led by Bain Capital
- Fast-moving startup environment with meaningful ownership
- Competitive compensation and early-stage equity
