Job Description
Machine Learning Engineer
Remote Role : Alpharetta Georgia 30009
W2 Candidates : with minimum validity of 12 months
Must have:
Need good clinical and claims knowledge
7+ years of healthcare exp
4+ years of experience in machine learning, NLP, or deep learning
They are working on AI evidence engine
focused on LLM
2 rounds of Zoom interview then offer.
What Youll Do
Develop, fine-tune, and optimize LLMs and modern deep learning models
Write high-quality prompts, instructions, and training examples to shape model behavior
Design, implement, and maintain instruction orchestration and evaluation workflows for LLM-based systems
Build and maintain training pipelines, datasets, and evaluation workflows
Design and execute functional and automated tests to validate AI outputs and system behavior
Analyze model performance, identify failure patterns (e.g., accuracy gaps, hallucinations, edge cases), and drive improvements
Collaborate with engineering and product teams (and review partner or vendor work) to deploy and iterate on AI features
Contribute to the ongoing maintenance and improvement of existing AI systems
What Were Looking For
7+ years of experience in machine learning, NLP, or deep learning
Hands-on experience with LLMs (GPT, LLaMA, Mistral, or similar) in applied or production contexts
Healthcare data experience is required, including working knowledge of:
Strong Python skills; experience with PyTorch or TensorFlow
Familiarity with HuggingFace tools and modern model-training workflows
Experience evaluating AI output quality, hallucination behavior, reliability, and consistency
Experience designing automated evaluation, regression testing, or benchmarking pipelines for AI systems
Ability to work with minimal direction, take ownership of problem areas, and operate effectively in ambiguous problem spaces
Excellent communication skills for writing prompts, instructions, technical documentation, and evaluation artifacts
Experience optimizing LLM and deep learning workloads on AWS, including model training, GPU utilization, and cost-efficient inference deployments