Svitla Systems Inc. is looking for a Senior AI Engineer for a full-time position (40 hours per week) in Ukraine.
Our client is an innovative startup developing an enterprise-grade AI agent platform that leverages cutting-edge AI research combined with pragmatic and reliable system engineering. Their mission is to build intelligent, cognitive AI agents capable of advanced reasoning, planning, and tool usage to solve complex business challenges.
About the role:
We are seeking a Senior AI Engineer to design and implement the cognitive core of AI agents within a distributed multi-agent platform. You will work at the intersection of state-of-the-art AI research and production-ready systems to build robust, scalable, and efficient AI solutions.
You will focus on advanced Retrieval-Augmented Generation (RAG) architectures, including graph-based reasoning, agentic systems, self-correction pipelines, and multimodal data processing. Your work will directly impact how the platform ingests, reasons over, and retrieves knowledge from large datasets, combining textual and visual information.
Overlap till 8 pm, ideally 9 pm UA time is required. Company Culture They’re building more than just cutting-edge AI — they’re building a culture that values clarity, purpose, and long-term impact. To thrive here, you’ll need to align with the following core principles: * Product-Centric Mindset: They care deeply about building products that deliver real value to non-technical users. Technology is a tool — not the goal. Engineers are expected to think in terms of user outcomes, simplicity, and usability. * Enterprise-Grade Pragmatism: While they leverage advanced AI research, our goal is to serve enterprise clients where reliability, security, and scalability are non-negotiable. They favor robust, testable, and maintainable code over experimental prototypes. * Holistic Systems Thinking: The platform is a distributed system of interconnected components. They expect engineers to understand and reason about the system as a whole — identifying dependencies, trade-offs, and potential failure modes before they become problems. * Deep Token Awareness: They treat tokens as the atomic units of cost, performance, and computation. You should understand tokenization from first principles, including how it affects latency, context windows, multi-agent coordination, and business costs.
Requirements * Professional experience building complex applications using AI orchestration frameworks such as LangChain or LlamaIndex. * Deep expertise in advanced RAG (Retrieval-Augmented Generation) architectures, well beyond basic vector search. * Corrective RAG: Implementing self-correction and validation loops to improve factual accuracy and reduce hallucinations. * AI Model Expertise: Solid understanding of cost-performance trade-offs between different model architectures. * Experience working with LLMs, SLMs, and models optimized for reasoning and efficiency (e.g., Language Reasoning Models, Mixture-of-Experts). * Strong systems engineering skills — able to build reliable, scalable, and maintainable solutions grounded in production needs. * Product mindset — building solutions that are easy to use and solve real business problems. * Ability to reason about tokens and their impact on performance, cost, and architecture (deep token awareness).
Nice-to-Have * LangGraph: Specific experience using LangGraph for building stateful, multi-agent workflows. * Graph-based RAG: Constructing and reasoning over knowledge graphs to uncover complex, multi-hop relationships. * Agentic RAG: Designing systems that reason, plan, and dynamically decide when and how to use tools during multi-step tasks. * Multimodal RAG: Experience building systems that ingest, embed, and reason over textual and visual data (e.g., PDFs with charts/images). * Hybrid Search: Implementing and tuning hybrid search systems combining keyword and semantic retrieval for precision.
Responsibilities * Design and implement core reasoning, planning, and memory architecture for AI agents. * Build systems enabling agents to reliably use external tools and knowledge bases. * Select, fine-tune, and optimize AI models to balance performance and cost. * Develop advanced Retrieval-Augmented Generation (RAG) systems including graph-based reasoning, agentic planning, corrective loops, and multimodal data processing (text + images). * Implement hybrid search systems combining keyword-based and semantic search for precise retrieval. * Work with AI orchestration frameworks such as LangChain, LlamaIndex, and preferably LangGraph. * Focus on building scalable, secure, and enterprise-grade solutions. * Deeply understand tokenization and its impact on system design, cost, and performance.
We offer * US and EU projects based on advanced technologies. * Competitive compensation based on skills and experience. * Annual performance appraisals. * Flexibility in workspace, either remote or in one of our development offices. * Comprehensive medical insurance including dental and massages. * Sport reimbursement program for onsite and online activities. * Support of a healthy lifestyle, compensation for running events. * Bonuses for recommendations of new employees. * Bonuses for article writing, public talks, other activities. * Personal loan budget available for long-term personnel. * 20 vacation days, 10 national holidays and sick leaves. * Maternity leave policy and family days off. * Full compensation for conferences, courses, English classes. * Free webinars, meetups and conferences organized by Svitla. * Gifts for New Year, anniversaries, children, and more. * Fun corporate celebrations and activities, regular lectures on various topics. * Awesome team, friendly and supportive community!
