AI Engineer

November 14, 2026

Remote

Middle

Albania, Georgia, Kazakhstan, Moldova, Montenegro, Serbia

Also worth knowing

Includes on-call duty

Description

The AI Engineer will own the quality, reliability and evolution of production AI systems, delivering end-to-end features and data-driven experiments. The stack includes TypeScript, Node.js, Cloudflare Workers, PostgreSQL, OpenSearch and Langfuse, with multiple LLM providers behind an AI gateway. The role also covers agentic workflows, tool integrations and RAG pipelines.

Responsibilities

Take ownership of large AI features and LLM scenarios and bring them to production within agreed timelines
Ensure quality, structure and predictability of AI user responses, achieving consistent structured output from different LLMs for similar requests
Integrate and coordinate multiple LLM providers through an AI gateway, managing model selection, fallback strategies and cost optimization
Design and orchestrate tool-based and agentic AI scenarios, including tool contracts, MCP tools, function calling interfaces and AI integrations with systems
Build and maintain multi-step LLM scenarios and RAG pipelines for indexing, embeddings, retrieval and reranking
Design and maintain production prompt systems with dynamic prompts, context injection and conditional logic
Manage prompts, LLM experiment releases and evaluation pipelines in Langfuse, including test datasets, LLM-as-a-judge and result evaluation
Run A/B tests of models and prompts, analyze results and present measurable impact of AI features and experiments
Investigate LLM chains via Langfuse traces, identify bottlenecks, optimize cost, latency and context window usage, and find causes of hallucinations and logical errors
Develop clean, scalable and maintainable TypeScript for the Node.js and Cloudflare Workers stack
Build reliable backend logic for AI systems, including request and response validation, error handling, fallback scenarios for invalid or low-quality model responses and reliable tool execution
Maintain code quality with tests, including checks for AI responses and fallback scenarios, conduct code reviews and follow engineering standards
Monitor AI system metrics, quality signals, latency and release status, resolve production incidents and document root causes
Document prompt design, model and routing choices, evaluation results and architectural decisions for handover to other specialists

Requirements

At least four years of experience in backend or fullstack development and commercial experience with TypeScript/Node.js
At least one year working with AI/LLM systems in production with real users, not only personal projects or experiments
Daily use of AI-assisted development tools such as Claude Code, Cursor or similar
Hands-on experience integrating LLM APIs from OpenAI, Anthropic, Google Gemini or similar in production, including structured output, function calling, streaming, token and cost control
Experience with agentic AI systems, multi-step orchestration, tool use scenarios or RAG pipelines with indexing, retrieval and reranking
Experience with LLM observability and evaluation tools such as Langfuse, LangSmith or similar
Experience with AI gateways and model routing such as OpenRouter, Cloudflare AI Gateway or similar
Strong understanding of relational databases, particularly PostgreSQL, and caching with Redis, KV or similar
Autonomy, ownership of results and high quality of engineering work
Nice to have: familiarity with evaluation frameworks including LLM-as-a-judge and RAGAS, as well as building test datasets
Nice to have: experience building MCP servers and clients or designing tool contracts for AI agents
Nice to have: experience with edge and serverless environments including Cloudflare Workers, KV, Durable Objects, Queues, R2 and Containers
Nice to have: experience with search engines and vector search such as OpenSearch, Elasticsearch or pgvector, as well as embeddings
Nice to have: experience with background job engines Hatchet, Temporal or Inngest for high-load LLM pipelines
Nice to have: experience with browser automation and browser agents using Playwright, CDP or browser-use
Nice to have: hands-on experience fine-tuning models for specialized tasks or strict compliance with JSON schemas
Nice to have: hands-on Python experience for data analysis, evaluation scripts or AI tooling

Conditions

Fully remote work from any country with a time window of approximately ±4 hours from CET
Unlimited paid vacation without counting days
National holidays are paid as days off
The company provides MacBooks to employees who need them for work
Flexible independent contractor agreement
Selection process includes a 40-minute recruiter screening
Technical interview in the selection process lasts 90 minutes
Final interview in the selection process lasts 60 minutes