AI Solution Architect

About the role

Prometa.ai is seeking an AI Solution Architect to lead the design and deployment of scalable, production‑ready AI infrastructures. The role combines deep knowledge of large language models, retrieval‑augmented generation, and autonomous agent workflows to turn experimental prototypes into enterprise‑grade solutions.

Key responsibilities

• Collaborate with business and IT teams to translate requirements into solution designs with defined inputs, outputs, success metrics and failure scenarios.
• Design end‑to‑end LLM solutions using RAG, tool‑calling and multi‑agent workflows, incorporating guardrails, fallback strategies and human‑in‑the‑loop checkpoints.
• Recommend model selection and serving architectures for on‑premise or closed‑network environments, considering vLLM, quantization, GPU and memory constraints.
• Build rapid proof‑of‑concepts, define technical roadmaps, effort estimates and production‑grade architectures.
• Integrate AI solutions with legacy, cloud‑native and enterprise systems, designing data ingestion, retrieval, reranking and vector or graph database layers.
• Establish evaluation and observability frameworks using golden datasets, LLM‑as‑a‑judge, safety metrics and regression testing.
• Embed security and compliance controls (PII protection, prompt‑injection defence, auditability) aligned with KVKK, BDDK, GDPR, MASAK and PCI‑DSS.
• Oversee containerised, GPU‑optimised deployments on Docker, Kubernetes or OpenShift, ensuring performance, fault tolerance and cost efficiency.
• Contribute to Prometa’s Orkestra product architecture, including telemetry schemas, SDK integration patterns and scoring pipelines.
• Lead customer workshops, demos and technical presentations, preparing architecture documents and RFP responses.
• Mentor engineers on architectural standards, conduct code reviews and provide technical guidance.

Required profile

• Bachelor’s degree in Computer Science, Software Engineering, AI or a related field.
• 6+ years of software, data or machine‑learning experience, with at least 2 years delivering LLM or ML systems in production.
• Proven track record designing and deploying an end‑to‑end GenAI system serving real traffic.
• Strong communication skills in Turkish and English, able to explain technical decisions to both technical and non‑technical stakeholders.

Required skills

• Advanced Python development, including FastAPI, REST APIs and microservices.
• Deep expertise in LLM engineering: prompt design, embeddings, chunking, vector databases, reranking, tool/function calling and hallucination mitigation.
• Experience with agent orchestration frameworks such as LangGraph, LlamaIndex, OpenAI Agents SDK or Semantic Kernel.
• Proficiency deploying open‑weight models in private infrastructure using Docker and Kubernetes; knowledge of vLLM, TGI, Ollama or NVIDIA Triton.
• Model quantization, inference optimisation, GPU and memory capacity planning.
• Familiarity with Turkish NLP, model and embedding selection, dataset preparation and quality evaluation.
• Building AI observability and platform products with OpenTelemetry, telemetry schemas, scoring pipelines and SDKs.
• Understanding of AI security practices: red‑team testing, prompt‑injection testing and OWASP Top 10 for LLM applications.
• Relevant cloud or enterprise architecture certifications (e.g., Azure AI Engineer, AWS Certified Machine Learning, Google Professional ML Engineer, TOGAF) are a plus.

What we offer

• Health insurance
• Meal card
• Transport allowance
• Training budget
• Gift card
• Birthday leave
• Happy hour events