LLM Engineer

Primary Responsibilities:

• End-to-end model development – design, train, fine-tune, and evaluate large-language-model (LLM) pipelines (GPT-class, Mistral, Llama 2/3, etc.) for scheduling, resource-allocation, and workflow-automation use-cases in hospitals.
• RAG & agents – build retrieval-augmented-generation and autonomous agent workflows that combine EHR, HR, IoT and historical case-length data to surface real-time recommendations.
• Prompt & instruction engineering – craft, test, and version prompts/instructions for accuracy, latency, and bias-mitigation under strict clinical-safety guidelines.
• Privacy-centric deployment – implement HIPAA / SOC 2-compliant pipelines (on-prem, VPC or secure PaaS) with guardrails, PHI redaction, audit logging and continuous monitoring.
• Performance tuning – optimize inference cost, token throughput, and latency across GPU clusters (NVIDIA A100/H100) and serverless endpoints.
• Collaboration – partner with data scientists, optimization researchers (OR-Tools, Hexaly), and product teams to ship LLM-powered features into our SaaS platform.
• Research-to-production – track new papers, evaluate SOTA methods (Mixture-of-Experts, LoRA, DPO, speculative decoding, tool-former agents) and run rapid POCs.
  

Requirements

• Proven experience in designing, developing, and deploying large language models, with a portfolio of past projects or contributions to LLM development.
• Advanced proficiency in Python and familiarity with libraries like TensorFlow, PyTorch, or Hugging Face Transformers
• Knowledge of tool automation, APIs, and vector databases
• Experience with modern LLM serving and inference frameworks,
• Hands-on experience with LangChain and LlamaIndex, enabling RAG applications and LLM orchestration.
• Strong software development skills with proficiency in Python. Experienced user of ML and data science libraries such as PyTorch, TensorFlow, Hugging Face Transformers, and scikit-learn.
• Familiarity with distributed computing, cloud infrastructure, and orchestration tools, such as Kubernetes, Apache Airflow (DAG), Docker, Conductor, Ray for LLM training and inference at scale is a plus.
• Ability to meaningfully present results of analyses in a clear and impactful manner, breaking down complex ML/LLM concepts for non-technical audiences.

Why Join Us?

At Opmed.ai, you’ll be part of a company that’s at the forefront of AI-driven healthcare solutions. We offer:

• A collaborative and intellectually stimulating environment.
• Opportunities to work on cutting-edge technology that directly impacts the healthcare industry.
• Competitive compensation, growth opportunities, and a fun, supportive team.