AI Architect

Craftware is a technology company of over 500 experts, empowering large organizations to solve complex business challenges with modern IT solutions - from sales systems and automation to data platforms and AI. We operate where technology must be reliable, secure, and scalable. We deliver end-to-end projects: from analysis and architecture through implementation to development and maintenance. We are a trusted partner of industry leaders such as Salesforce, Veeva, UiPath, and Databricks.

Model: hybrid - two or three times a week

Engagement: full-time

About the project

We are looking for an experienced Platform Architect (AI) to design and scale AI solutions across the organization. You will translate business needs into production-ready architectures spanning computer vision, LLMs, document processing, and data-driven automation.

The first project is a visual similarity search system for automated verification of in-store display compliance against planograms, based on object detection, image embeddings, and vector retrieval.

Responsibilities

Architecture & Technology Selection

• Design end-to-end AI/ML architectures, from PoC to production.

• Select models, frameworks, and tools for each use case.

• Run build vs. buy analyses, including cost modelling and vendor lock-in risk.

• Ensure solutions are scalable, secure, cost-efficient, and compliant with GDPR and the EU AI Act.

MLOps & Production Readiness

• Define MLOps/LLMOps standards: retraining triggers, drift and quality monitoring, model and data versioning, rollback strategy.

• Work with MLOps engineers on implementation.

• Integrate AI components with existing systems and data infrastructure.

Security & Compliance

• Act as the technical contact for the AI Security team and involve them in key design decisions.

• Ensure access control, encryption, and protection against adversarial inputs (e.g. spoofed images, prompt injection).

• Verify new third-party integrations against existing security approvals.

Visual Similarity Search Project

• Design the pipeline: product detection → visual embedding, OCR, and category classification → hybrid vector query → planogram matching.

• Select models and components (e.g. RF-DETR, DINOv2) based on accuracy, cost, licensing, and delivery time.

• Define category-specific retrieval strategies (e.g. text-driven for cigarettes, visual-driven for bottles).

• Decide between self-hosted GPU on AWS and managed services (Roboflow Inference, Pinecone).

• Define the API contract with the backend and RELEX integration, including handling of low-confidence matches.

• Set quality thresholds (precision/recall per category, false-positive rate) with the Product Owner.

• Verify license compliance of components (e.g. Apache vs. AGPL-3.0).

• Design the feedback loop for adding field-verified image crops to the index.

• Define data security and retention requirements with AI Security, including GDPR-relevant store photos.

Leadership & Communication

• Provide architectural code reviews and guidance to AI/ML Engineers.

• Mentor engineers on AI architecture patterns and best practices.

• Maintain architecture documentation, diagrams, and decision records.

• Track emerging AI technologies (foundation models, vector databases, agentic frameworks, RAG, fine-tuning) and assess their fit.

Requirements

Required Skills & Experience

• 5+ years in software architecture, including 2–3 years in AI/ML systems.

• Strong knowledge of machine learning, deep learning, and LLM architectures.

• Hands-on experience with production computer vision pipelines (object detection, embeddings, vector/hybrid search).

• Familiarity with DETR/YOLO-family models and Roboflow or a similar tool.

• Experience with vector databases (Pinecone or equivalent), RAG, and agentic frameworks.

• Experience designing multi-signal retrieval or ranking systems.

• Strong proficiency in Python and ML frameworks (PyTorch, TensorFlow, Hugging Face).

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

• Working knowledge of AWS ML/GPU services (SageMaker, EC2 GPU, Bedrock).

• Ability to evaluate self-hosted vs. managed inference costs.

• Understanding of open-source licensing (AGPL, Apache, MIT) and its business impact.

• Pragmatic decision-making under time and budget constraints.

• Experience leading or mentoring engineering teams.

• Excellent communication skills, including with non-technical stakeholders.

Nice to Have

• Experience in retail (product recognition, planograms, display compliance).

What we offer:

• B2B contract

• Fully remote service delivery

• Broad range of projects (internal, international) - genuine variety of clients and tasks

• Budget for skills development and certifications as part of the collaboration

• Regular collaboration reviews and discussion of project scope

• Additional benefits available as part of the collaboration

• Networking and team-building events for project teams

Tech stack

• Python
• PyTorch
• AWS
• vector databases
• MLOps