AI / LLM Engineer (Claude, Prompt Architecture)

FluidLabs is building production-grade AI systems and is looking for an engineer experienced in designing and operating LLM-based architectures.

The role requires practical experience with Claude (Opus / Sonnet), prompt architecture, context management strategies, and structured LLM workflows integrated into backend services.

Key responsibilities:

• Design and implement production-grade LLM workflows using Claude (Opus / Sonnet)
• Architect structured prompt systems (system/user/tool separation, multi-step flows)
• Design context management strategies (long-context handling, memory layering)
• Reduce hallucinations through prompt strategy, evaluation loops, and guardrails
• Integrate Claude via API into backend services (Node.js / TypeScript or Python)
• Build RAG pipelines (embeddings, vector search, context injection)
• Design LLM-driven automation (agents, tool calling, workflow orchestration)
• Evaluate and improve response quality through structured testing
• Optimize performance, latency, and cost efficiency
• Collaborate with backend and architecture teams to embed AI into real systems
   

Must-have technical skills:

• 4–5 years of backend engineering experience (Node.js or Python)
• Hands-on production experience with Claude (Anthropic API)
• Strong understanding of:
  • Prompt architecture
  • Context construction and management
  • Multi-step reasoning flows
  • LLM limitations and hallucination patterns
     
• Experience integrating LLM APIs into backend systems
• Understanding of RAG concepts and embeddings
• Experience working with structured JSON outputs and tool-calling patterns
• Ability to explain architectural trade-offs clearly
   

Nice to have:

• Experience building AI agents or autonomous workflows
• Familiarity with vector databases (pgvector, Pinecone, Weaviate)
• Experience with AWS (Lambda, API Gateway)
• Experience in multi-tenant SaaS systems
• Understanding of AI safety and PII handling

What we offer:

• Opportunity to shape a new DocuSign-focused platform from the ground up
• High level of technical ownership and architectural influence
• Enterprise-level projects with real architectural and scalability challenges
• Long-term collaboration on meaningful projects
• A passionate team with a healthy, positive culture
• Enterprise-level projects with real technical challenges
• Room to grow — both professionally and personally
• Referral bonuses for successful candidate recommendations
• Paid holidays and time off
   

Work terms:

• Working hours: 11:00–20:00 (Kyiv time)
• Vacation: 24 paid days per year
• Trial period: 3 months