AI Platform Architect

Quick Facts

• Rate: 210–240/250 PLN/hour (invoices)

• Work model: Mostly remote; visit Warsaw office 2x/week

• Employment: Full-time

• Sector/Project context: Petrol/Convenience Retail

• Project: AI Visual Similarity Search to automatically verify product display compliance with store planograms

Description

You will design and architect an end-to-end AI/ML platform for Visual Similarity Search. The role focuses on building computer-vision and vector-search pipelines, selecting models and technologies, defining API contracts and quality gates, and overseeing monitoring, model versioning, and rollback. You will collaborate with AI/ML engineers, DevOps, Product, and AI Security to ensure GDPR and EU AI Act compliance.

Responsibilities

• Architect AI/ML systems end-to-end

• Design computer vision pipelines and vector search (including retrieval/ranking concepts)

• Select models/technologies; optimize costs (build vs. buy); integrate with backend systems and RELEX

• Define API contracts and quality thresholds

• Provide architectural oversight of the AI/ML engineer

• Design monitoring, model versioning, and rollback mechanisms

• Analyze open-source licenses

• Work with AI Security on safety and compliance (GDPR, EU AI Act)

• Document architecture and mentor the engineering team

Requirements

• 5+ years of software architecture experience; at least 2–3 years focused on AI/ML systems

• Strong fundamentals in machine learning and deep learning (including modern NLP/LLM architectures)

• Hands-on experience with cloud AI/ML platforms (AWS/Azure/GCP)

• Experience with MLOps (e.g., MLflow, Kubeflow, SageMaker) and ML CI/CD for pipelines

• Fluency in Python; experience with PyTorch/TensorFlow and Hugging Face

• Experience with vector databases, RAG pipelines, and agentic AI frameworks

• Knowledge of data architecture, APIs, and integration patterns

• Ability to explain complex technical topics to non-technical stakeholders

• Experience leading technical teams or mentoring engineers

• Production experience with computer vision pipelines (object detection, embedding models, vector/hybrid search)

• Knowledge of DETR/YOLO model ecosystem and Roboflow (or similar) for data/model training

• Experience with vector databases (e.g., Pinecone or equivalent)

• Ability to evaluate cost trade-offs for self-hosted GPU inference vs managed/SaaS inference

• Understanding of open-source license constraints (AGPL, Apache, MIT) and business consequences

• Experience designing multi-signal retrieval/ranking systems

• Practical AWS ML/GPU knowledge (SageMaker, EC2 GPU, Bedrock) for architectural discussions with DevOps

Benefits

• High flexibility for remote work

• Long-term stable employment

• Chance to participate in key projects for a large, stable-sector organization

• Attractive compensation aligned with skills and experience

• Benefits: Lux Med, MultiSport