Jr. AI/ML Engineer at Astiva Health – Orange, California

About Us:

Astiva Health, Inc., located in Orange, CA is a premier health plan provider specializing in Medicare and HMO services. With a focus on delivering comprehensive care tailored to the needs of our diverse community, we prioritize accessibility, affordability, and quality in all aspects of our services. Join us in our mission to transform healthcare delivery and make a meaningful difference in the lives of our members.

SUMMARY:

We are seeking a skilled and adaptable Junior AI/ML Engineer to join our fast-moving team building impactful AI solutions in healthcare. Our work focuses on extracting and interpreting data from unstructured medical documents, improving clinical coding accuracy, streamlining administrative processes, and enhancing patient outreach.

Projects will evolve rapidly, from fine-tuning large language models (LLMs) on specialized medical PDFs, to optimizing OCR pipelines in Azure, and new challenges emerge regularly. This role suits someone who thrives in ambiguity, enjoys hands-on model development, and wants to directly influence healthcare delivery through applied AI/ML.

ESSENTIAL DUTIES AND RESPONSIBILITIES include the following:

• Design, fine-tune, and optimize large language models (LLMs) and multimodal models for healthcare-specific NLP tasks, such as information extraction, classification, and summarization from clinical documents (e.g., medical charts, patient files, scanned forms).
• Develop and improve document understanding pipelines, including fine-tuning OCR / layout-aware models (especially in cloud environments like Azure AI, Azure Foundry) to handle real-world variability in medical forms, handwriting, and scanned PDFs.
• Build and iterate on end-to-end ML solutions that transform unstructured healthcare data into structured, actionable insights
• Collaborate closely with clinicians, product managers, data annotators, and engineers to define problems, curate/annotate datasets, evaluate model performance against clinical and business metrics, and iterate quickly.
• Deploy models into production environments (cloud-based inference, batch processing, or API endpoints) with attention to latency, cost, scalability, and healthcare compliance considerations (HIPAA, data privacy).
• Stay current with advancements in LLMs, vision-language models, efficient fine-tuning techniques (LoRA/QLoRA, PEFT), RAG, multimodal AI, and domain-specific healthcare AI research.
• Contribute to a culture of rapid prototyping, rigorous evaluation, and continuous improvement in a dynamic project landscape where priorities can shift based on new opportunities or stakeholder needs.
• Other duties as assigned

OTHER SKILLS and ABILITIES:

• Hands-on fine-tuning experience with LLMs (e.g., Hugging Face, OpenAI fine-tuning, Azure Foundry), even if limited to small-scale or academic projects
• Exposure to cloud ML platforms (Azure ML, AWS SageMaker, or GCP)
• Familiarity with RAG architectures and retrieval-based grounding
• Experience with NLP libraries (spaCy, Hugging Face Transformers, NLTK)
• Introductory experience with weak supervision or noisy-label learning
• Interest in healthcare or biomedical NLP
• Curiosity about knowledge graphs, ontologies, or structured prediction
• Familiarity with secure data handling practices
• Willingness and ability to learn workflows for sensitive or regulated data (e.g., HIPAA-covered healthcare data), including privacy-aware data handling and secure ML workflow

EXPERIENCE:

• Bachelor’s Degree in related field
• 1–2 years of experience in machine learning, applied NLP, or software engineering
• Demonstrated some experience training or fine-tuning ML models, not just using APIs
• Ability to collaborate with senior engineers and domain experts and incorporate feedback

BENEFITS:

• 401(k)
• Dental Insurance
• Health Insurance
• Life Insurance
• Vision Insurance
• Paid Time Off
• Free catered lunches

REQUIRED TECHNICAL SKILLS:

• Proficiency in Python and familiarity with common ML frameworks (e.g., PyTorch, TensorFlow, scikit-learn)
• Experience applying NLP techniques to unstructured text
• Hands-on experience working with LLMs, including:
• Prompt design and iteration
• Using pre-trained models for classification or extraction tasks
• Foundational understanding of model fine-tuning, such as:
• Fine-tuning transformer models or LLMs for classification or information extraction
• Adapting existing training scripts or examples to new datasets
• Familiarity with model evaluation metrics (precision, recall, F1) and basic error analysis
• Experience working with labeled datasets and annotation outputs, including reviewing label quality
• Understanding of common ML problem types, including binary and multi-label classification
• Awareness of model bias, label noise, and false positives, with the ability to discuss tradeoffs and mitigation strategies
• Basic understanding of production ML workflows (versioning, reproducibility, monitoring concepts)