Computer Vision Engineer

Mill; San Bruno (CA); On-site; USD 220,000 - 250,000; Full time

Job summary

Mill is seeking a Computer Vision Engineer to advance the CV technology behind Mill Commercial, turning food waste streams into actionable intelligence for large kitchen operators.

You will own modeling, training infra, and cloud-side evaluation harnesses, while guiding data-groundtruth workflows and model deployment decisions. This IC role emphasizes hands-on work with fine-tuning, MLOps pipelines, and end-to-end system awareness.

Qualifications

• Strong fundamentals in computer vision — segmentation, detection, classification, tracking.
• Experience with modern ML approaches including VLMs, LLMs, and foundation models.
• Experience designing robust evaluation metrics and ML eval harnesses.
• Product shipping experience with models in production, drift handling and latency budgeting.
• Ability to justify tooling decisions with data and explicit criteria.

Responsibilities

• Train and evaluate segmentation, classification, and mass-estimation models for the Mill Commercial camera pipeline.
• Optimize edge models for production performance and scale the ML pipeline end-to-end.
• Create and curate datasets per customer/vertical to meet accuracy targets across food types and environments.
• Analyze failure cases and drive data/modeled decisions to close accuracy gaps.
• Build annotation tooling and ground-truth generation workflows, including foundation-model-assisted labeling.
• Collaborate with MLOps and edge engineers on training practices, versioning, and deployment tradeoffs.

Skills

Computer vision

Segmentation

Detection

Classification

Tracking

Python

PyTorch

OpenCV

VLMs

LLMs

Foundation models

Agentic systems

Tools

Python

PyTorch

OpenCV

LLM frameworks

Weights & Biases

Job description

About the Role

We're hiring a Computer Vision Engineer to work on the CV technology behind Mill Commercial — the computer vision and agentic systems that turn a stream of food waste into operational intelligence for commercial kitchens. Mill Commercial integrates a camera into our high-capacity food recycler; models identify and quantify food scraps, and our pipeline turns that signal into procurement and operational guidance for large food service operators.

You'll join a small, capable team, owning the modeling and training infrastructure that powers our CV technology. You will design the cloud-side evaluation harness to determine if edge models meet production targets and build the ground-truth workflows to support them. This is a hands-on IC role for someone who brings deep computer vision fundamentals to fine-tuning models, building MLOps pipelines, and establishing a methodical approach to managing system complexity.

What You'll Do

• Train and evaluate segmentation, classification, and mass-estimation models for the Mill Commercial camera pipeline — from prompting foundation models to fine-tuning ConvNets and VLMs.
• Optimize edge models for production performance, and operationalize and scale the ML pipeline with model lineage tracking end to end.
• Create and curate datasets per customer/vertical — more customized, purpose-driven data — to support accuracy targets across food types, kitchen environments, and deployment configurations.
• Analyze failure cases systematically — unfamiliar food classes, novel kitchen environments, challenging lighting and clutter conditions — and drive the data and modeling decisions that close accuracy gaps.
• Build annotation tooling and ground-truth generation workflows, including foundation-model-assisted labeling, to keep pace with model iteration.
• Bring a methodical approach and strong opinions, backed by experience, to the modeling and evaluation decisions you own — and partner with the team's MLOps and edge engineers on training practices, versioning, and deployment tradeoffs as they come up.

What We're Looking For

• Strong fundamentals in computer vision and deep learning — segmentation, detection, classification, tracking — deep enough to make informed architecture calls.
• Fluency with modern ML approaches — VLMs, LLMs, foundation models, and agentic systems — alongside classical deep learning. You know when to fine-tune a ConvNet, when to prompt a VLM, and when to wire up an agent, and you understand the practical realities of putting any of them into a product.
• Experience evaluating ML models rigorously — designing metrics, building eval harnesses, and using results to drive product decisions rather than just publish a number.
• Product shipping experience — you've taken a model to production and dealt with what comes after (drift, edge cases, latency budgets), not just to a benchmark.
• Bias for action — you'd rather ship a good-enough experiment and learn from it than wait for the perfect plan.
• Experience making build‑vs‑buy or tooling decisions backed by data or a clear rubric, not just instinct — you can show your work on how you got there.
• Clear, direct communication — you can explain tradeoffs to non-technical stakeholders, push back honestly when you disagree, and write docs that others can follow.
• Genuine interest in applying AI to food waste reduction and sustainability. This is a mission-driven product and we want people who care about the mission.
• Software skills: Python, PyTorch, OpenCV. Experience with LLM and agent frameworks.

Nice to Have

• Experience with video understanding (temporal consistency, tracking, video segmentation)
• Experience with MLOps tooling (Weights & Biases, MLflow, SageMaker, ClearML, or equivalents)
• Hardware / IoT product experience, particularly with computer vision and cameras for embedded systems

The estimated base salary range for this position is $220-250K, which does not include the value of benefits or a potential equity grant. A wide range of factors are considered in making compensation decisions, including but not limited to skill sets, market conditions, experience and training, licensure and certifications, and business and organizational needs. At Mill, it is not typical for an individual to be hired at or near the top of the range for their role.