Senior Computer Vision Engineer

Senior Computer Vision Engineer | Join the team behind a $3.2B smart-home exit, now taking on food waste | Bay Area (Hybrid) | $225K–$300K baseWe've partnered with a Bay Area hardware company on a mission to prevent waste, starting with food. Founded by the team behind one of the most successful connected home products of the last decade, they build smart systems and infrastructure for homes, businesses and municipalities that turn food scraps into a valuable resource instead of sending them to landfill. Tens of thousands of their home recyclers are already in kitchens, diverting large volumes of food waste every year, and they are now launching a first-of-its-kind commercial product: an end-to-end system for managing, understanding and preventing food waste in settings like grocery, restaurants and food service. It is a real machine in a real environment, and the computer vision work sits right at the centre of the next chapter.This is a rare opportunity to own the computer vision behind that commercial product. The system puts a camera inside a high-capacity food recycler: models identify and quantify what passes through, and the pipeline turns that signal into procurement and operational guidance for large food-service operators. You'll join a small, capable team and own the modelling and training infrastructure that powers it, designing the cloud-side evaluation harness that decides whether edge models meet production targets and building the ground-truth workflows underneath. It is a hands-on IC role for an engineer who has already put vision models onto a device and lived with what happened next.The RoleAs a Computer Vision Engineer, you will:Train and evaluate segmentation, classification and mass-estimation models for the camera pipeline, from prompting foundation models to fine-tuning ConvNets and VLMs.Optimise models to run on the constrained hardware inside the machine, and operationalise the ML pipeline end to end with model lineage tracked throughout.Create and curate purpose-built datasets per customer and vertical to hold accuracy across food types, kitchen environments and deployment configurations.Analyse failure cases systematically, from unfamiliar food classes to novel kitchens and difficult lighting and clutter, and drive the data and modelling decisions that close the gaps.Build annotation tooling and ground-truth workflows, including foundation-model-assisted labelling, to keep pace with model iteration.Partner with the team's MLOps and edge engineers on training practices, versioning and deployment tradeoffs as they come up.About YouYou have taken computer vision models into a shipped product and owned what came after: drift, edge cases, latency and memory budgets, and the failures that only appear once real units are in the field. A benchmark result is not the same thing.You have strong fundamentals in computer vision and deep learning, across segmentation, detection, classification and tracking, deep enough to make informed architecture calls.You are fluent with modern approaches such as VLMs, LLMs, foundation models and agentic systems alongside classical deep learning, and you know when to fine-tune a ConvNet, when to prompt a VLM and when to wire up an agent, including the practical realities of putting any of them into a product.You evaluate models rigorously: designing metrics, building eval harnesses, and using the results to drive product decisions rather than to publish a number.You have worked on cameras and vision running on embedded or IoT hardware, and you understand the tradeoffs that come with a fixed compute budget inside a physical machine.You have a bias for action: you make build-versus-buy and tooling calls backed by data or a clear rubric, show your working, explain tradeoffs to non-technical stakeholders, and push back honestly when you disagree.You are fluent in Python, PyTorch and OpenCV, with experience using LLM and agent frameworks, and you care about applying AI to food-waste reduction.One of these backgrounds fits you: a computer vision engineer who has shipped models into a real product, or an ML engineer working across classical CV and modern foundation models who has owned evaluation and MLOps as well as modelling.Nice to have: video understanding such as temporal consistency, tracking and video segmentation; MLOps tooling such as Weights & Biases, MLflow, SageMaker or ClearML; and time at an early-stage company where you owned more than one part of the system.You are based in the San Francisco Bay Area and comfortable with a hybrid working pattern.The role pays a competitive base with flexibility for the right background, plus a potential equity grant. The team is moving quickly with this hire. If you're interested in owning the computer vision that turns a stream of food waste into operational intelligence, on a product built to keep food out of landfill, apply now or send your CV directly to will@inventurerecruitment.com.