End-to-End ML Engineer for Real-World Vision & Edge

Blissway Inc.; Denver, Northern (CO, KY); Hybrid; USD 120,000 - 180,000; Full time

Benefits offered by this job

Relocation bonus

Health coverage

401(k) matching

Life & disability insurance

Equity package

4 weeks PTO

12 weeks parental leave

Sabbatical after 5 years

Job summary

Blissway Inc. is hiring a Machine Learning Engineer to own the full model lifecycle from raw sensor data to production inference.

You will deploy and monitor models on real roads, handling data curation, training, and continuous improvement across vision tasks like detection, segmentation, embeddings and re-identification. We process 11 million images daily and operate across cloud and roadside hardware, balancing speed, power, and accuracy.

Qualifications

• 2 to 6 years of software engineering with focus on machine learning and/or computer vision.
• End-to-end experience: trained, deployed, monitored, and iterated in production.
• Comfort across both software engineering fundamentals and hands-on ML.
• Experience with modern computer vision depth; know when classical methods beat heavy models.

Responsibilities

• Own the whole pipeline: from raw sensor data to production inference, including collection, dataset curation, training, deployment, monitoring, and iteration.
• Work with real hardware in the field; test ideas on real roads.
• Process 11 million images daily with detection, segmentation, embeddings, and re-identification.
• Build models spanning classical CV to custom SOTA; select appropriate tools and train your own when needed.
• Balance edge and cloud compute for fast, power-efficient inference without sacrificing accuracy.
• Collaborate in a small, autonomous team delivering production-ready ML shipping to roads.

Skills

Python

TypeScript

Job description

Blissway Inc. is hiring a Machine Learning Engineer to own the full model lifecycle from raw sensor data to production inference.

You will deploy and monitor models on real roads, handling data curation, training, and continuous improvement across vision tasks like detection, segmentation, embeddings and re-identification. We process 11 million images daily and operate across cloud and roadside hardware, balancing speed, power, and accuracy.