machine learning engineer

Ciklum is a custom product engineering company that supports multinational organizations and scaling startups in solving complex business challenges. The company engineers technology to redefine industries and improve lives, leveraging a global team of developers, consultants, analysts, and product owners.

• Own the architecture decision process for the AI track, including agent framework selection, orchestration topology, and model routing policy;
• Define the canonical agent blueprint and ensure all production agents adhere to these standards;
• Design and manage the evaluation system, including ground-truth corpus design, accuracy harnesses, and regression gates;
• Lead AI safety and security initiatives, including adversarial testing and defense against prompt injection;
• Build high-risk components personally, such as the agent SDK core and evaluation harness;
• Set LLM engineering policies, including token-economics, cost attribution, and fallback strategies;
• Direct knowledge-graph and RAG strategy for secure, multi-tenant environments;
• Define and enforce MLOps/LLMOps standards, including observability and eval-gated CI/CD;
• Lead and grow the AI engineering pod, managing technical performance and delivery quality;
• Design and operate the "Agent Factory" model to standardize and accelerate agent delivery;
• Mentor senior engineers and contribute to practice-wide standards;
• Conduct technical assessments for new AI engineering hires.

• 12+ Years in software/AI engineering with 3+ years building production LLM-powered systems, including agentic or multi-agent platforms;
• Expert-level Python and deep experience with production agent frameworks (LangGraph, Claude Agent SDK);
• Proven experience designing evaluation systems that gated real releases, including LLM-as-judge calibration and hallucination measurement;
• Hands-on experience with AI security, including prompt-injection defense, tool-call authorization, and adversarial testing;
• Ability to author and defend architecture decisions for systems involving durable execution (Temporal), multi-tenancy, and vector retrieval (Neo4j, PGVector);
• Production experience with LLMOps, including tracing, cost observability, and eval-gated CI/CD;
• Experience leading engineering teams of 4+ through milestone-driven deliveries and setting technical standards;
• Strong communication skills for negotiating with client CTOs and architecture boards;
• Nice to have: Familiarity with Commercial Real Estate workflows, compliance exposure (SOC 2, GDPR, FedRAMP), experience in platform productization, and community standing through publications or conference talks.

• Medical insurance;
• Mental health support;
• Financial and legal consultations;
• Access to internal events, meetups, conferences, and workshops;
• Udemy access and company-paid certifications;
• Opportunities for internal mobility and professional growth.