Senior Staff Software Architecture-AI Algorithm

Job Description Summary

Develop and optimize LLM, VLM, embedding, and audio multimodal models based on
open-source foundation models. Build GUI-recognition and automated-control agents for medical
post-processing software and medical devices, meeting product design, performance, and
compliance-validation requirements. This role does not involve large-scale foundation-model
pretraining.

GE HealthCare is a leading global medical technology and digital solutions innovator. Our purpose is to create a world where healthcare has no limits. Unlock your ambition, turn ideas into world-changing realities, and join an organization where every voice makes a difference, and every difference builds a healthier world.

Job Description

Multimodal Fine-Tuning & Agent Development

• Fine-tune LLM, VLM, embedding, and audio models using PEFT methods including LoRA/QLoRA, SFT, and DPO.

• Deliver GUI recognition, UI-element localization, instruction understanding, task planning, and tool-calling capabilities for device and software automation.

Data, Evaluation & Traceability

• Build multimodal datasets from software screenshots, workflows, device commands, and audio interactions; perform collection, cleaning, annotation, and augmentation.

• Establish automated evaluation for recognition accuracy, instruction-following success, hallucination rate, and robustness; maintain versioning and experiment records for R&D traceability.

Model Optimization & Deployment

• Apply distillation, pruning, and INT4/INT8 quantization (PTQ/QAT) to reduce model size, memory, and compute cost while preserving accuracy.

• Optimize inference through KV cache, operator fusion, and memory reuse; convert and deploy models locally using TensorRT and ONNX Runtime to meet latency and throughput targets.

Integration, Validation & Research

• Integrate VLM interface recognition, LLM planning, embedding retrieval, and ASR understanding into an automated agent loop with engineering teams.

• Support model verification, performance validation, technical documentation, risk analysis, boundary-case analysis, and evaluation of emerging VLM, Agent, GUI, and lightweight-model technologies.

Qualifications

Education

Master's degree or above in Computer Science, Artificial Intelligence, Pattern Recognition, Biomedical Engineering, or a related field.

Fundamental Knowledge

• Strong understanding of Transformer architectures, LLMs, VLMs, and PyTorch.

• Hands-on end-to-end experience with SFT and parameter-efficient fine-tuning such as LoRA/QLoRA.

Required Technical Skills

• Experience with VLMs (for example, LLaVA), GUI/screenshot visual understanding, and automated-control tasks.

• Knowledge of embedding models, RAG, and ASR model tuning.

• Knowledge of distillation, quantization, pruning, TensorRT, and ONNX Runtime.

• Proficiency in Python and Linux; ability to independently build datasets, train and evaluate models, and complete model conversion and deployment.

· Strong knowledge of the core training methods for conventional convolutional neural networks (CNNs), with hands-on end-to-end CNN training experience; able to independently perform computer-vision data processing, model development and training, hyperparameter tuning, and performance evaluation, and capable of diagnosing and resolving practical CV problems through systematic error analysis and solution implementation.

Preferred Qualifications

• Experience in medical devices, medical-imaging post-processing software, medical-device AI, or medical-device software quality systems.

• Practical experience with agent development, tool calling, and orchestration frameworks such as LangChain or LangGraph.

• Experience optimizing models for embedded or compute-constrained local devices; familiarity with C++, inference SDKs, or vector databases.

Professional Competencies

• Strong focus on experiment traceability, model boundaries, and risk management.

• Sound engineering judgment to balance accuracy, latency, and hardware cost.

• Effective cross-functional collaboration with device software, Product, and Quality teams

Inclusion and Diversity

GE HealthCare is an Equal Opportunity Employer where inclusion matters. Employment decisions are made without regard to race, color, religion, national or ethnic origin, sex, sexual orientation, gender identity or expression, age, disability, protected veteran status or other characteristics protected by law.

We expect all employees to live and breathe our behaviors: to act with humility and build trust; lead with transparency; deliver with focus, and drive ownership – always with unyielding integrity.

Our total rewards are designed to unlock your ambition by giving you the boost and flexibility you need to turn your ideas into world-changing realities. Our salary and benefits are everything you’d expect from an organization with global strength and scale, and you’ll be surrounded by career opportunities in a culture that fosters care, collaboration and support.

Additional Information