This is not a research or experimentation role. Prompt engineering in our company is system design. You will design reliable, scalable, production ready LLM workflows that power real customer facing capabilities inside a complex distributed platform.
You will work closely with backend engineers, product managers, and architects to integrate LLM based intelligence into core business flows.
Location: Kfar Saba, Israel.
Reporting to: AI Team Lead.
Roles and Responsibilities
Design, implement, and continuously improve prompts for LLM driven product features.
Architect and develop backend services in Python.
Integrate LLM APIs such as OpenAI, Anthropic, and AWS Bedrock into production systems.
Implement structured output enforcement, schema validation, and response normalization.
Design robust error handling, fallback strategies, retries, and resiliency mechanisms.
Optimize latency, token usage, throughput, and API cost efficiency.
Build evaluation frameworks and quality control pipelines for AI outputs.
Collaborate with Product and Engineering teams to deliver AI features end to end within us.
Knowledge and Experience
4+ years of backend development experience with strong proficiency in Python.
Proven hands on experience building and shipping GenAI powered features to production.
Strong experience with Python GenAI frameworks such as LangChain, LangGraph, Strands, or similar orchestration frameworks.
Experience integrating LLM APIs into live distributed systems.
Experience implementing structured outputs, validation layers, and guardrails.
Familiarity with evaluation frameworks and LLM quality measurement techniques.
Experience building RESTful APIs.
Strong understanding of clean architecture, scalability, and production best practices.
Advantages
Experience designing and implementing RAG pipelines.
Experience with MCP servers, A2A architectures, or multi agent systems.
Experience working with embeddings and vector databases.
Proficiency in TypeScript (Node.js or ReactJS).
Experience with AI observability, monitoring, and evaluation tooling.
