ROIFORCIO GmbH (Austria, R&D in Ukraine) is looking for an AI RAG DevOps / Developer the Pitch Avatar product team. We are looking for a hands-on practitioner to lead the migration and optimization of our RAG infrastructure. Full remote work.
Skills: result-oriented, proactive, deep understanding of vector search algorithms, ability to solve complex performance bottlenecks, practical rather than theoretical mindset.
Requirements:
β’ 3+ years of experience in Backend Development (Python or Golang);
β’ 1+ year of practical experience building and maintaining RAG (Retrieval-Augmented Generation) systems in production (not just Pet projects);
β’ Deep experience with Vector Databases: Zilliz Cloud / Milvus (priority), or Weaviate/Qdrant/Pinecone;
β’ Proven track record of optimizing RAG performance: experience reducing ingestion time for large files and achieving low-latency search (< 3s);
β’ Strong experience with AWS infrastructure; knowledge of Azure and GCP is a strong plus (multi-cloud strategy);
β’ Experience in DevOps: containerization, deployment, and managing cloud resources for AI workloads;
β’ Experience handling Multimodal content: parsing and indexing PDF, XLS, MS Word, PPTX, TXT, MP3, MP4;
β’ Experience with Structured Data Integration: connecting LLMs to relational DBs and SQL tables (Hybrid Search);
β’ Experience creating MVPs from scratch and customizing solutions;
β’ Understanding of MCP (Model Context Protocol) and Local LLM deployment is a plus;
β’ English at Intermediate level (B1+).
Duties:
β’ RAG Migration & Architecture: Lead the migration from AWS OpenSearch to Zilliz Cloud (Milvus) or a similar high-performance Enterprise solution to solve current cost and speed issues.
β’ Performance Engineering: Optimize the ingestion pipeline to ensure content addition (up to 10MB) takes no more than 10 seconds.
β’ Latency Optimization: Ensure Avatar responses (LLM + RAG) are delivered in under 2-3 seconds while maintaining high relevance and utilizing full context.
β’ Multimodal Data Processing: Develop robust pipelines for parsing and indexing diverse file formats (documents, audio, video) in multiple languages.
β’ Hybrid Search Implementation: Implement logic to query both unstructured vector data and structured relational data (SQL/Tables) simultaneously.
β’ Infrastructure Scalability: Prepare the architecture for future multi-cloud support (Azure, GCP) and integration of external RAG sources.
β’ DevOps & Deployment: Manage the deployment of AI services, ensuring stability and handling "noisy neighbor" issues in a multi-tenant environment.
β’ Innovation: Support the integration of future technologies like MCP and local LLMs.
Technological Stack: Python/Golang, Zilliz Cloud (Milvus), AWS (Azure/GCP planned), OpenAI, LangChain/LlamaIndex, SQL, Docker, Kubernetes.
Work:
Contract of Private Entrepreneur with Austrian company
Remote work
Please see product and try first: pitchavatar.com
Distributed Team
Required skills experience:
Vector DBs (Zilliz/Milvus/Weaviate) 1+ year
Backend (Python/Go) 3+ years
DevOps/Cloud (AWS) 2+ years
RAG Architecture 1+ year
Required languages: English B2
About Pitch Avatar
Pitch Avatar is a startup that is changing the approach to training, presentations and sales. We create smart avatars that speak for you, teach beginners, conduct demos and do not get tired. We utilize advanced RAG infrastructure to make our avatars knowledgeable, but we need you to make them faster and smarter!
