Founding ML Engineer, Computer Vision (Object Detection)

OpenKrill; United States; On-site; USD 200,000 - 260,000; Full time

Job summary

Openkrill is seeking a founding machine learning engineer to steer computer vision for a resale marketplace. You will build and productionize models that identify items from images with high accuracy to support pricing and customer decisions.

You will own from research to production, including latency, cost, and reliability, while communicating capabilities and tradeoffs to stakeholders. This fully remote role is open to North America-based candidates.

Qualifications

• 5+ years of applied computer vision experience.
• Experience with fine-grained or instance-level classification.
• Proficiency with PyTorch or TensorFlow, including fine-tuning and deploying vision transformers or CNNs.
• Experience building evaluation frameworks measuring real-world model performance.
• Ability to communicate tradeoffs to non-technical stakeholders.
• Experience with active learning or HITL labeling, and low-latency model APIs.

Responsibilities

• Design computer vision architecture for fine-grained item identification, building on foundation models and adapting them to a specific catalog.
• Define accuracy standards by category and develop calibrated confidence scores so the product can recognize when it is uncertain.
• Build feedback loops that use model errors to guide future labeling and model improvement.
• Set priorities between expanding category coverage and improving accuracy in existing categories.
• Own the path from research to production, including model serving latency, cost, and reliability.
• Communicate model capabilities and limitations to technical and non‑technical stakeholders.

Skills

Applied computer vision

Production deployment

Fine-grained classification

PyTorch

TensorFlow

Evaluation frameworks

Stakeholder communication

Active learning

Job description

ABOUT THE ROLE

As the founding machine learning engineer, you will set the technical direction for computer vision in an early‑stage resale marketplace. You will build and productionize models that identify items from images with the accuracy and confidence needed to support trustworthy pricing and customer decisions.

WHAT YOU'LL DO

• Design computer vision architecture for fine‑grained item identification, building on foundation models and adapting them to a specific catalog.
• Define accuracy standards by category and develop calibrated confidence scores so the product can recognize when it is uncertain.
• Build feedback loops that use model errors to guide future labeling and model improvement.
• Set priorities between expanding category coverage and improving accuracy in existing categories.
• Own the path from research to production, including model serving latency, cost, and reliability.
• Communicate model capabilities and limitations to technical and non‑technical stakeholders.

WHAT WE'RE LOOKING FOR

• At least 5 years of applied computer vision experience, including a system shipped to production at meaningful scale.
• Experience with fine‑grained or instance‑level classification, where distinguishing similar items matters.
• Strong PyTorch or TensorFlow skills, with production experience fine‑tuning and deploying vision transformers or CNNs.
• Experience building evaluation frameworks that measure real‑world model performance and improvement.
• Comfort serving as the senior technical voice on an open‑ended problem without an established playbook, and explaining tradeoffs to non‑technical stakeholders.
• Experience with active learning or human‑in‑the‑loop labeling, low‑latency model APIs, or an early‑stage startup is valuable.

COMPENSATION & BENEFITS

Annual salary range: $200,000 to $260,000.

LOCATION

Fully remote within North America.