Computer Vision Engineer

Assembler AI; Montreal (administrative region); On-site; CAD 120,000 - 180,000; Full time

Benefits offered by this job

Competitive salary and stock options

Health and wellness spending account

Latest MacBook and modern tooling

Opportunity to grow with the company

Job summary

Assembler AI in Montreal is seeking a founding computer vision engineer to work directly with the Head of AI on perception models behind our products, owning datasets and evaluation workflows to improve field performance.

This hands‑on role focuses on data-driven progress, with a strong emphasis on real footage analysis, end‑to‑end experiments, and collaboration with an annotation team to drive improvements on live production lines in Canada.

Qualifications

• Bachelor's in computer science, engineering, mathematics, or related field (or equivalent practical experience). Master’s in computer vision, ML or related area welcome but not required.
• 2+ years of hands-on applied computer vision experience, ideally in industry or on systems deployed beyond a research environment.
• Experience training and deploying vision models on real-world data end-to-end.
• Strong Python and PyTorch skills.
• Solid understanding of evaluation metrics and their relation to production performance.
• Comfortable reviewing video footage and understanding what cameras see.
• Proficient in French and English.
• Based in Canada.

Responsibilities

• Train, evaluate, and improve vision models on real production video.
• Own training datasets: sourcing data, defining label schemas, and reviewing annotation quality.
• Design evaluations that reflect real deployment conditions and report results clearly.
• Investigate model failures directly on footage and identify root causes.
• Work with annotation team to target data most likely to improve performance.
• Run experiments end to end: from question, to training runs, to actionable results.

Skills

Python

PyTorch

OpenCV

NumPy

Pandas

Linux

AWS

GCP

Git

Education

Bachelor's degree in Computer Science or related field

Master's in computer vision, ML welcome but not required

Tools

Git

Linux

AWS

GCP

CUDA

PyTorch

OpenCV

Job description

The role

We\'re hiring a founding computer vision engineer. You\'ll work directly with the Head of AI on the perception models behind our products, and take ownership of the datasets and evaluation work that make them perform in the field.This is an applied, hands‑on role. Our hardest problems aren\'t about picking the newest architecture , they\'re about building the right training data, evaluating honestly enough to know what actually improved, and spending real time in the footage to understand what a model is getting wrong. If you like problems where the answer is in the data rather than the paper, you\'ll do well here.

What you\'ll do

• Train, evaluate, and improve vision models on real production video


• Own our training datasets: sourcing and selecting data, defining label schemas, and reviewing annotation quality


• Design evaluations that reflect real deployment conditions, and report results clearly and honestly


• Investigate model failures directly on footage and identify their root causes


• Work with our annotation team to target the data most likely to improve performance


• Run experiments end to end: from question, to training runs, to a result someone can act on

Requirements

• Bachelor\'s in computer science, engineering, mathematics, or a related field ( or equivalent practical experience). A Master\'s in computer vision, machine learning, or a related area is welcome but not required


• 2+ years of hands‑on applied computer vision experience, ideally in industry or on systems deployed beyond a research environment


• Y\'ve trained and deployed a vision model on real‑world data end to end, including the messy parts


• Strong Python and PyTorch


• Solid understanding of evaluation: train/test splits, precision/recall tradeoffs, and why a strong validation score doesn\'t always mean a good model in production


• Comfortable spending significant time reviewing video footage; understanding what the camera actually sees is a real part of this job


• Willing to say \"I don't know\" and \"I think I got that wrong.\" We\'d rather hear it early


• Professional working proficiency in French and English


• Based in Canada

Technical environment

Required:

• Git and a standard branch/review workflow


• Linux — comfortable working entirely over SSH on remote GPU machines: bash, background jobs, services, reading logs


• Cloud — hands‑on experience with AWS and/or GCP, including compute, storage, and managing your own training environments


• PyTorch and CUDA — GPU‑based training, checkpoint management, and experiment tracking.


• OpenCV and ffmpeg — decoding, cropping, and processing video at scale


• NumPy, pandas, scikit‑learn, and standard annotation data formats

Nice to have:

• Experience with video and temporal models, not just single images


• Object tracking and multi‑object association


• Model export and inference optimization


• Experience with annotation platforms and running an annotation workflow


• Exposure to manufacturing, robotics, or industrial inspection

You\'ll be the second person on the AI team, with direct ownership of a core part of the product and a very short path from your work to something running on a live production line.

• Competitive salary and stock options


• Health and wellness spending account, from $500 to $1,000 annually


• Latest MacBook and modern sales tooling


• Opportunity to grow as the company scales