Senior/Staff MLOps Engineer - Computer Vision

Attis; Boston (MA); On-site; USD 160,000 - 200,000; Full time

Job summary

Attis, a Boston-based Series B startup, is seeking an experienced MLOps Engineer to own the production ML stack for a live computer vision product.

You will migrate serverless prototypes to enterprise-grade Kubernetes infrastructure, architect cloud-native pipelines on GCP, and optimize edge vs cloud inference across devices including embedded probes. On-site four days a week in Boston.

Qualifications

• 5+ years of professional software/ML engineering experience, with significant tenure owning live production ML systems.
• Proven background taking multi-stage computer vision models from prototype to high-volume production serving.
• Strong production cloud engineering expertise within GCP (GKE, Cloud Run, Vertex AI, Cloud Functions, Pub/Sub) and Docker.

Responsibilities

• Own the end-to-end multi-stage computer vision inference pipeline across latency, throughput, reliability, and cost.
• Re-architect the serving stack onto containerized, auto-scaling GPU/CPU clusters using Triton, TorchServe, or Cloud Run.
• Build automated deployment pipelines with canary rollouts, shadow traffic evaluation, and rapid rollback mechanisms.
• Establish production monitoring infrastructure for data drift detection and real-world failure triage.
• Develop closed-loop active learning workflows, automated labeling pipelines, dataset versioning, and offline regression tests.
• Evaluate on-device versus cloud inference trade-offs for mobile capture probes using CoreML, TFLite, or ExecuTorch.
• Serve as the primary cloud architecture authority across GCP, Terraform IaC, and container standards.

Skills

ML engineering

Computer vision

GCP expertise

Model serving

Tools

Docker

Kubernetes

GKE

TensorRT

Job description

$160,000 – $200,000 Base Salary + Annual Bonus + Early-Stage Equity

The Company

• Series B startup automating vehicle servicing through robotics and computer vision systems.
• Backed by tier-one automotive giants who serve simultaneously as strategic investors and enterprise launch customers.
• Deploying proprietary physical AI hardware and optical inspection tools directly into commercial service bays nationwide.
• High-growth engineering hub operating out of a dedicated facility in Boston, MA.

Why Join?

• Complete technical ownership of the production machine learning platform for a live, commercially deployed optical inspection product.
• Lead the migration of an early-stage serverless vision prototype to a high-throughput, industrial-grade inference engine.
• Work directly with proprietary, un-scrapable real-world image datasets collected from operational service lanes.
• Highly collaborative, low-ego engineering culture with direct access to executive leadership and rapid execution cycles.
• Prime technical role bridging computer vision algorithms, cloud architecture, and embedded hardware capture probes.

The Role

A technical, hands-on leadership role where you will:

• Own the end-to-end multi-stage computer vision inference pipeline (segmentation and classification) across latency, throughput, reliability, and cost.
• Re-architect the serving stack off serverless functions onto containerized, auto-scaling GPU/CPU clusters using Triton, TorchServe, or Cloud Run.
• Build robust automated deployment pipelines featuring canary rollouts, shadow traffic evaluation, and rapid rollback mechanisms.
• Establish production monitoring infrastructure for continuous data drift detection, technician override tracking, and real-world failure triage.
• Develop closed-loop active learning workflows, automated labeling pipelines, dataset versioning, and rigorous offline regression test suites.
• Evaluate on-device versus cloud inference trade-offs for mobile capture probes utilizing CoreML, TFLite, or ExecuTorch.
• Serve as the primary cloud architecture authority across GCP, establishing Terraform infrastructure-as-code and container standards.

The Essential Requirements

• 5+ years of professional software/ML engineering experience, with significant tenure owning live production ML systems.
• Proven background taking multi-stage computer vision models (segmentation and classification) from prototype to high-volume production serving.
• Strong production cloud engineering expertise within GCP (GKE, Cloud Run, Vertex AI, Cloud Functions, Pub/Sub) and Docker.
• Demonstrated mastery of high-performance model serving frameworks such as Triton Inference Server, TorchServe, TensorRT, or ONNX Runtime.
• Located in the Greater Boston area with willingness to work on-site 4 days per week.
• Existing US Work Authorization (US Citizen, Permanent Resident, or active H-1B transfer).

What Will Make You Stand Out

• Hands-on experience deploying edge vision models onto mobile devices or embedded hardware via CoreML or TensorFlow Lite.
• Prior engineering work with industrial inspection, physical sensors, optical probes, or noisy automotive diagnostic datasets.
• Track record migrating early-stage serverless prototypes to enterprise-grade Kubernetes infrastructure.

No terminology in this advert is intended to discriminate on the grounds of age, sex, race, religion or belief, disability, pregnancy and maternity, marriage and civil partnership, sexual orientation, gender, and/or gender reassignment, and we confirm that we are happy to accept applications from anyone for this role.

Keywords for Search (SEO): MLOps Engineer, Machine Learning Operations, Computer Vision, Triton Inference Server, GCP, GKE, TorchServe, Kubernetes, PyTorch, TensorRT, Docker, Deep Learning, TensorFlow, Model Serving, Cloud Run