Senior LLM Engineer – (Training, Evaluation, Production Deployment and Agentic Rag)

We are seeking a highly skilled Data Scientist to join our innovative SaaS-based insurance mining platform. We have experienced record growth and are excited to grow the team. The ideal candidate will have industrial real world production expertise in fine-tuning large language models (LLMs), deep learning, and retrieval-augmented generation (RAG). This role requires hands-on experience with containerized workflows (e.g., Docker), Azure cloud services, and LLM optimization tools such as Unsloth, Lora and Qlora. You will be responsible for building, optimizing, and deploying models with large context windows to drive value for our insurance clients. This is direct hire role only no third parties or agencies. Candidates from Argentina, Pakistan and India encouraged to apply All Candidates will be required to take an automated test, followed by white board session. Candidate must be fluent to the highest level in written and spoken English language. This is a fully remote position. In order to be considered for this role: Candidate must have at least 4 years of experience with title ML Engineer or Data Scientist. Candidate must have 2 years continuous employment with the same employer Candidates must have completed a Stem related degree (not have one in progress) Recent grads are NOT a fit for this role. Key Responsibilities: 1. Model Development & Fine-tuning: Fine-tune and deploy large-scale LLMs (e.g., GPT, OPT, Llama, Falcon,Qwen) to extract insights from structured and unstructured insurance data (e.g., policy documents, claims data). Candidate must have experience with agentic AI with preference on Langchain/Langsmith Leverage transfer learning and parameter-efficient fine-tuning (LoRA, PEFT) to optimize performance for specific tasks, such as document summarization and claims processing. Implement large-context-window models to handle long insurance documents, improving query accuracy in complex data extractions. 2. Retrieval-Augmented Generation (RAG) Systems: Develop RAG pipelines to enhance LLM performance by integrating with external knowledge sources (e.g., Pinecone, FAISS, Azure Cognitive Search). Implement query encoding and embedding models for more accurate and contextual RAG queries. Optimize embedding-based document search for large insurance databases using vector databases. 3. Containerization & Cloud Deployment: Design and manage containerized environments using Docker and Docker Compose for reproducible training and inference workflows. Deploy and orchestrate LLMs in Azure Kubernetes Service (AKS) and Azure Machine Learning (AML). Implement distributed training workflows for large LLMs using Azure Databricks, DeepSpeed, or Ray. 4. Model Performance & Monitoring: Monitor and improve the performance of LLM-based models (e.g., F1-score, perplexity, inference time) with tools such as Unsloth for efficient model evaluation and experimentation. Implement error analysis tools and automated retraining pipelines using MLOps best practices. Optimize for cost-efficiency and scalability in Azure cloud environments. 5. Collaboration & Stakeholder Communication: Collaborate with data engineers, product managers, and business stakeholders to understand requirements and deliver machine learning solutions tailored to client needs. Communicate complex findings and insights effectively through dashboards, reports, and presentations. Required Qualifications: Education: Bachelor’s or Master’s degree in Data Science, Computer Science, Machine Learning, Applied Mathematics, or a related field. A PhD is a plus. Experience: At least two years as data scientist, NLP engineer or MLOPS engineer in last role. 4+ years of experience in building and fine-tuning large language models (LLMs). Experience with deep learning frameworks (e.g., PyTorch, TensorFlow) and libraries like Hugging Face Transformers. Hands-on experience with retrieval-augmented generation (RAG), vector databases (e.g., Pinecone, Weaviate, FAISS), and semantic search. Strong experience in containerized workflows using Docker and Kubernetes. Experience with Azure cloud services, including Azure Machine Learning (AML), Azure Blob Storage, and AKS. Strong Linux experience, shell scripting Technical Skills: Expert in Python. Familiarity with Unsloth or similar model evaluation frameworks for large LLM fine-tuning. Strong experience in embedding models (e.g., SentenceTransformers) and distributed training. Knowledge of MLOps frameworks (e.g., MLflow, Azure Pipelines) for versioning, monitoring, and model retraining. Preferred Qualifications: Experience working with insurance workflows (e.g., underwriting, claims analysis). Familiarity with long-document models (e.g., Llama2-Long, BigBird, or Memorizing Transformers, Qwen large models). Experience with low-rank fine-tuning (LoRA) and quantization techniques for efficient model deployment. Knowledge of Azure OpenAI Services for large-scale NLP applications. Soft Skills: Strong problem-solving and analytical thinking. Ability to work independently and manage multiple projects simultaneously. Effective communication and collaboration with cross-functional teams. About LineSlip Solutions: LineSlip has created a unique data visualization platform purpose-built for the commercial insurance industry that uses technology to automatically extract and organize data previously locked in binders, policies, proposals, and other insurance documents. With LineSlip Solutions, users can easily visualize data, automate reporting, and make smarter, more informed business decisions that affect the bottom line. We are actively working with some of the country’s most recognizable companies.