Senior GenAI Engineer with German at Accesa RaRo

, , Romania -

Full Time

Start Date

Immediate

Expiry Date

16 Jun, 26

Salary

0.0

Posted On

18 Mar, 26

Experience

5 year(s) or above

Remote Job

Yes

Telecommute

Yes

Sponsor Visa

Skills

Python, Prompt Engineering, LLM, RAG Systems, Vector Databases, LangChain, Context Optimization, AI Safety, APIs, Deployment, Observability, Pandas, NumPy, Scikit-learn, German, Semantic Search

Industry

IT Services and IT Consulting

Description

Company Description Company Description Accesa is a leading technology company headquartered in Cluj-Napoca, with offices in Oradea and 20 years of experience in turning business challenges into opportunities and growth. A value-driven organisation, it has established itself as a partner of choice for major brands in Retail, Manufacturing, Finance, and Banking. It covers the complete digital evolution journey of its customers, from ideation and requirements setup to software development and managed services solutions. With more than 1,200 IT professionals, Accesa also has a fast-growing footprint, establishing itself as an employer of choice for IT professionals who are passionate about problem-solving through technology. Coming together in strong tech teams with a customer-centric approach, they enable businesses to grow, delivering value for our clients, partners, industry, and community. Job Description One of our clients operates prominently in the financial sector, where we enhance operations across their extensive network of 150,000 workstations and support a workforce of 4,500 employees. Our IT solutions ensure streamlined processes and heightened security, enabling them to maintain leadership in financial technology. Responsibilities Prompt Engineering & Enhancement : Design, secure, extend, and customize system prompts (e.g., chat history, instructions, rules); evaluate the impact of prompt changes and continuously improve LLM usage scenarios while analyzing risks and benefits. RAG Systems Development: Design and optimize retrieval pipelines, including document chunking, embedding generation, and ranking strategies. Build and maintain solutions using vector databases (e.g., FAISS, Weaviate, Pinecone) or Postgres with pgvector, to enable accurate and context-aware responses. Context Management & Optimization: Implement advanced context window optimization, memory strategies, and token efficiency techniques to ensure scalable and cost-effective LLM interactions. LLM Evaluation, Monitoring & Tracking: Evaluate new LLM releases, monitor response quality and consistency, and build automated evaluation pipelines to assess performance across use cases. AI Safety & Guardrails Implementation: Design and implement prompt injection protection, output filtering, and policy enforcement mechanisms to ensure safe and compliant AI behavior in enterprise environments. System Integration & APIs: Develop and expose AI capabilities via scalable APIs, ensuring proper integration into existing systems and workflows. Deployment, Serving & Scalability: Design and manage serving layers for LLM applications, addressing scaling strategies, latency considerations, and reliability in both cloud and on-prem environments Observability & Reliability: Implement logging, monitoring, and alerting mechanisms to ensure visibility into system performance, usage patterns, and failures. Python Development & Prototyping: Develop prototype and production-level Python code to support AI features, ensuring reliability and maintainability aligned with product requirements. Knowledge Sharing & Research: Stay up to date with the latest techniques in reasoning, chaining, and context optimization; contribute to AI Labs community discussions and share findings with the Data Science team. Qualifications Language mandatory requirement: Strong German language skills (B2 level or higher) required Must have: 4+ years of experience with Python and its data science ecosystem (Pandas, NumPy, Scikit-learn) Proven experience in prompt engineering and LLM behavior analysis Extensive hands-on experience with LLM integration frameworks such as LangChain or similar. Experience designing retrieval pipelines, including chunking, embedding generation, and ranking strategies Hands-on experience with vector databases (PG Vector, FAISS, Pinecone, Weaviate, etc.) Strong understanding of embeddings and semantic search Experience with context window optimization, memory strategies, and token efficiency techniques Experience implementing prompt injection protection, output filtering, and policy enforcement mechanisms Experience building and exposing APIs for AI systems Experience with serving LLM applications, including scaling and latency optimization Familiarity with deployment in cloud and/or on-premise environments Experience with logging, monitoring, and system observability Experience building automated evaluation pipelines for LLM systems Nice to have: Experience with Java or modern backend frameworks (e.g., Spring) for enterprise system integration Experience with frontend frameworks (e.g., Angular) for building or integrating AI-driven user interfaces Familiarity with LLM observability tools or advanced evaluation frameworks Experience with real-time or streaming inference systems As a people-first organisation, we believe diversity strengthens our teams and drives innovation. All employment decisions are based on merit, skills, and performance, without discriminating based on any personal characteristic. This reinforces our commitment to providing an inclusive and respectful workplace. Additional Information At Accesa you can Enjoy our holistic benefits program that covers the four pillars that we believe come together to support our wellbeing, covering social, physical, emotional wellbeing, as well as work-life fusion. Physical Wellbeing: Our wellbeing program includes medical benefits, gym support, and personalised fitness options for an active lifestyle, complemented by team events and the Healthy Habits Club. Work-Life Fusion: In very dynamic industries such as IT, the line between our professional and personal lives can quickly become blurred. Having a one-size-fits-one approach gives us the flexibility to define the work-life dynamic that works for us. Emotional Wellbeing: We believe that to maintain our overall health, we need to invest in our mental wellbeing just as much as we do in our physical health, social connections or in achieving work-life balance. Social Wellbeing: As a growing community in a hybrid environment, we want to ensure we remain connected not just by the great work we do every day but through our passions and interests. Department: Core Technologies | Data & AI

Responsibilities

The role involves designing, securing, and enhancing system prompts, alongside developing and optimizing Retrieval Augmented Generation (RAG) systems using vector databases for context-aware responses. Key duties also include implementing AI safety guardrails, building scalable APIs for AI capabilities, and managing LLM serving layers for deployment and scalability.