UA - SSr. LLM Engineer - Job2307

Job Summary

We are looking for a talented and innovative Large Language Model (LLM) Engineer with specialized experience in Retrieval Augmented Generation (RAG) to join our dynamic team. The ideal candidate will be instrumental in the development, optimization, and deployment of RAGbased LLM solutions that meet various business needs. This role requires a deep understanding of Data Science, Machine Learning (ML), Natural Language Processing (NLP), and the fundamentals of LLMs. The successful candidate will have hands-on experience with agent based prompting techniques such as LangChain, APIdriven development architectures for AI products, usage of vector databases, and information retrieval pipelines to extract actionable insights. Additionally, strong listening skills and an ability to understand business requirements are essential for designing effective solutions tailored to stakeholder needs.

Job Responsibilities

• Designing RAG Solutions: Develop robust RAGbased LLM solutions that enhance data retrieval processes while ensuring high quality output aligned with business objectives.
• Prompt Engineering: Craft effective prompts that efficiently guide LLMs towards generating outputs that fulfill specific business requirements; iterate on prompt designs based on performance metrics.
• AI Product Development: Design, develop, and refine AI products based on user feedback and stakeholder requirements; ensure alignment with program priorities through agile methodologies.
• Support ML Engineering Teams: Collaborate closely with ML engineering teams during the deployment phase of APIs in production environments; assist in troubleshooting issues related to model performance or integration challenges.
• Documentation Creation: Document Generative AI use cases comprehensively; create manuals detailing workflows, methodologies employed in solution design, and best practices adopted throughout the process.
• Retrieval Augmented Generation Maintenance: Develop and maintain RAG concepts following established data science principles; ensure adherence to industry standards while fostering innovation.
• Pipeline Optimization: Ensure efficient operation of data pipelines and processes according to company security policies; promote Responsible AI practices within all project phases.
• Collaboration & Communication: Work closely with cross functional teams including product managers, data scientists, software engineers, and stakeholders to align goals effectively; communicate technical concepts clearly across diverse audiences.

Basic Qualifications

• A minimum of 3 years' experience working with Python for machine learning applications along with expertise in AWS Cloud Computing services.
• Familiarity with AWS Bedrock's suite of LLM models along with associated prompting strategies tailored for optimal performance.
• At least 1 year of direct experience working specifically within NLP frameworks/LLMs focusing on Retrieval Augmented Generation models.
• Proficiency in Python programming language utilized extensively for data science projects involving AI development.
• Familiarity with agent based frameworks like LangChain which support advanced interactions between users/models.
• Experience leveraging AI training frameworks such as TensorFlow or PyTorch alongside Hugging Face Transformers library for model building/training tasks.
• Demonstrated expertise in prompt engineering techniques aimed at finetuning large language models tailored toward specific applications or industries.
• Knowledgeable about vector databases including practical experience implementing embedding techniques necessary for enhanced information retrieval capabilities.

Preferred Skills

• Experience utilizing orchestration tools designed specifically for AI product development (e.g., ArgoCD).
• Ability to create detailed diagrams illustrating data flows/solutions which can effectively communicate complex ideas/processes among team members/internal stakeholders alike.
• Prior experience deploying UI/frontend components related directly back into existing large language model systems would be advantageous.