Senior Computer Vision / ML Engineer

About Bright RiverBright River provides Visual Content services to leading e-commerce and retail companies worldwide. Our 2D and 3D visual experts deliver tens of thousands of product images daily using our proprietary platform and applications. With offices across the world, we're building the future of automated image editing.Our AI team is a dedicated group of computer vision and deep learning experts developing ML-powered Editing Automation that combines cutting-edge AI with human expertise to unlock new capabilities for our customers.The OpportunityWorking directly with leadership to define and execute our Computer Vision and Machine Learning strategy. This is not a staff augmentation role—you'll have genuine ownership over critical AI products that directly impact our customers and thousands of images processed daily.Your MissionTransform Bright River's editing services through state-of-the-art Computer Vision and Deep Learning. You'll build, deploy, and maintain production ML systems that augment our skilled visual experts, creating a powerful synergy between human creativity and AI precision.What You'll Own:Core ResponsibilitiesModel Development & ResearchDesign, train, and deploy custom computer vision models for image editing workflows (segmentation, object detection, quality enhancement, style transfer)Fine-tune foundation models (Stable Diffusion, SAM, CLIP) on proprietary datasetsExperiment with novel architectures and stay current with SOTA research (transformers, diffusion models, GANs)Benchmark open-source models vs. custom solutions (cost, quality, IP ownership)Production ML SystemsBuild end-to-end ML pipelines: data preprocessing → training → evaluation → deploymentImplement distributed training on multi-GPU infrastructure (PyTorch DDP, Ray)Containerize and deploy models to Kubernetes with CI/CD automationSet up experiment tracking (MLflow, Weights & Biases) and model versioningModel Monitoring & IterationImplement drift detection, performance monitoring, and alertingAnalyze failure modes and work with data team on continuous improvementA/B test model variants against baseline (designer feedback, processing time)Document model cards (architecture, training data, limitations, performance)Strategic CollaborationCross-Functional IntegrationPartner with Engineering team to integrate AI models into editing workflowsCoordinate with Operations team on SLAs, quality standards, and production incidentsWork with visual experts (designers, QA team) to understand editing requirements and gather feedbackCollaborate with leadership to define AI roadmap and product prioritiesTechnical LeadershipMentor junior engineers on CV best practices and ML engineeringConduct code/model reviews to maintain quality standardsPresent findings and demos to stakeholders (weekly model reviews, quarterly planning)Contribute to AI team's knowledge base and documentationYour ImpactWithin your first year, you will:Ship 2-3 production AI models that generate immediate business impact.Establish ML infrastructure (training pipelines, deployment automation, monitoring) used by the entire AI teamInfluence product strategy by identifying high-value AI opportunities through data analysisContinue building foundations for Bright River's AI-first editing platform serving global customersRequirementsMust-Have Technical SkillsComputer Vision Expertise5+ years professional experience in CV/ML roles at commercial companiesDeep understanding of CV fundamentals: CNNs, Vision Transformers, segmentation techniques (semantic/instance), object detectionProven track record: shipped 3+ CV models to production environmentsExpert in PyTorch or TensorFlow (can implement research papers from scratch)Production ML EngineeringHands-on experience with full ML lifecycle: training, deployment, monitoringProficiency in Docker, Kubernetes (or equivalent container orchestration)CI/CD implementation for ML systems (version control, automated testing, deployment pipelines)Cloud platforms experience: Azure, AWS, or GCP (GPU workloads, storage, compute)Software EngineeringStrong Python engineering: clean, testable, maintainable codeFamiliarity with software design patterns and Clean Architecture principlesExperience with distributed systems and asynchronous processingComfortable with Git workflows, PR reviews, and collaborative developmentRequired Education & ExperienceMSc in Computer Science, Computer Vision, Machine Learning, or related field (or equivalent practical experience)5+ years in CV/ML engineering roles (not research-only positions)Fluent in English (written and verbal)Essential Soft SkillsResults-oriented problem solver: You ship working solutions, not just experimentsStrong communicator: Can explain complex ML concepts to non-technical stakeholdersCollaborative mindset: Thrive in cross-functional teams with diverse cultural backgroundsSelf-directed: Comfortable with ambiguity, can define technical approaches independentlyPragmatic researcher: Balance staying current with SOTA vs. delivering business valueNice-to-Have (Stand Out from the Crowd)Advanced Technical SkillsExperience implementing custom neural network architectures from scratch (low-level PyTorch/TensorFlow)Hands-on with generative models: diffusion models (Stable Diffusion, DALL-E), GANs, VAEsModel optimization expertise: quantization (INT8, ONNX), TensorRT, distillationMLOps tooling: Kubeflow, Ray, Seldon, TensorFlow ServingDomain KnowledgeFamiliarity with Adobe Creative Cloud (Photoshop, Lightroom) or image editing workflowsUnderstanding of 3D rendering, CGI, or virtual photographyExperience in e-commerce, retail, or visual content industriesKnowledge of image quality assessment and perceptual metricsLeadership & ResearchPublished research (papers, blog posts) or open-source contributions in CV/MLExperience mentoring junior engineers or leading technical initiativesTrack record of presenting at conferences or internal tech talksPhD in Computer Vision or Machine Learning (if coupled with production experience)What We OfferCompensation & BenefitsCompetitive salary based on experience (40-hour workweek)Holiday allowance, and holidays off per yearTravel expense reimbursement for office commutesWork EnvironmentFlexible working hours: Core hours for collaboration, flexibility outsideHybrid work model: Haarlem office, remote otherwiseLearning budget: Conferences, courses, books, certificationsGrowth & ImpactFounding team member role: Shape AI strategy, not just execute ordersDirect leadership access: Work closely with CTO and product leadershipInternational exposure: Collaborate with teams across 5 cities, 4 countries, 3 continentsCulture & TeamPassionate professionals: Team that works hard and celebrates wins togetherDiverse, inclusive environment: Colleagues from 10+ countries and backgroundsInnovation-driven: Encouraged to experiment, fail fast, and learn continuouslyCustomer impact: See your models processing thousands of images daily for global brandsAbout You (Ideal Candidate Profile)You are a builder who ships. You love the thrill of seeing your model process its first production image, not just publishing a paper. You have strong opinions on PyTorch vs. TensorFlow (and can articulate why), but you're pragmatic—if fine-tuning Stable Diffusion solves the problem in 2 weeks vs. training from scratch in 6 months, you choose the former.You're excited by the intersection of research and engineering. You read arXiv papers on Sunday mornings, but on Monday you're debugging why your model segmentation fails on transparent product images. You understand that "99% accuracy" means nothing without latency SLAs, cost per inference, and edge case handling.You're a collaborative technologist. You don't gatekeep ML knowledge—you teach designers how to interpret model confidence scores. You write documentation before someone asks for it. When a model fails, you don't blame "bad data"—you work with the data team to fix labeling workflows.Most importantly, you're motivated by impact. The idea that your solutions are improving the results of our customers, and being seen by millions of end customers energizes you more than hitting a benchmark on an academic dataset.