Relevant Experience & Education Highlights

Expertise in generative diffusion models, LLM alignment, and multimodal signal processing aligns strongly with developing IP-aware guardrails for Adobe Firefly's generative systems.

1. Multimodal IP-Aware Generative Modeling

Developed generative techniques integrating constraints into model behavior, mirroring needs for proactive IP guidance in Firefly.

• Developed generative AI diffusion model for visualizing stroke areas from microwave signals at a medical technology startup, achieving image quality comparable to MRI and CT scans.
• Implemented weapons detection using GAN-generated datasets at a security systems company, reducing development costs by half.
• Created proof-of-concept vehicle damage detection leading to a granted patent at a multinational engineering firm.

2. Vision-Language & Multimodal Reasoning

Advanced multimodal systems combining vision, language, and signal processing for semantic understanding and safety-aware inference.

• Performed end-to-end signal processing and ML-based stroke prediction on microwave transceiver networks at a medical technology startup, improving accuracy above 70% across 100+ patients.
• Conducted LLM prompt engineering and evaluation using models like ChatGPT and Claude for shopping Q&A at a major e-commerce company.
• Led research on AI for cancer prediction and planning, resulting in a high-citation journal paper acceptance during academic role.

3. Research & Technical Depth

Demonstrated depth in fine-tuning, alignment, and large-scale ML with PyTorch, directly supporting Firefly's generative model requirements.

• Evaluated and implemented systems for LLM application alignment and fine-tuning at a major e-commerce company.
• Leveraged customer data for size recommendation systems impacting billions of users, simplifying architectures to reduce retraining from weeks to days at a Fortune 500 technology company.
• Implemented state-of-the-art deep learning algorithms for corrosion classification from image analysis as research assistant at a top-tier engineering school.

4. Rapid Scientific Experimentation

Drove rigorous experiments evaluating trade-offs in generative pipelines, aligning with needs for new benchmarks and failure mode analysis.

• Utilized PyTorch, TensorFlow, and diffusion models in medical imaging and predictive modeling projects across multiple roles.
• Applied transfer learning, neural networks, and computer vision techniques including OpenCV for multimodal R&D at various technology and research positions.
• Conducted literature reviews and bi-weekly collaborations with research scientists on deep learning implementations.

Requirements & Candidate Alignment