Catalin Caldararu

I spend half of my time leading and developing an Applied AI intellectual property (IP) portfolio consisting of 25 differentiated Hybrid AI, On-device AI and Agentic AI software product concepts. I own the portfolio's strategy, prioritization, and lifecycle, and I contribute directly, working with a core team of five inventors and multiple legal teams. Each product can be applied across one of nine domains: Orchestration, Efficiency, Resilience, Security, Privacy, Accessibility, Personalization, Field Operations, and Agent-Native SDLC. Key responsibilities:

• Initiate and develop patentable AI product concepts aligned with Dell's AI Strategy, targeting competitive differentiation, defensible IP positions, and potential growth vectors.
• Drive the patent development process from opportunity discovery through iterative refinement, including business‑case development, use‑case and persona definition, technical design, and visual prototyping.
• Define portfolio and product strategies by identifying underserved technological whitespace and novel positioning opportunities, aligning with the company's strategy and discontinuing IP initiatives when necessary.
• Manage the IP backlog using Kanban and EBM, applying Kaizen principles to continuously refine, reprioritize and increase the long-term strategic value of each product concept.
• Create detailed technical architectures that enable independent execution by engineering teams.

I spend half of my time as a core contributor to the SLM and ASR services that power privacy‑first, on‑device Conversational AI for Dell AI PCs. These services enable enterprises to run AI workflows without cloud dependency and are part of Dell's AI PC platform, featured as a flagship AI initiative at CES, DTW, and GTC. Key responsibilities:

• Work in a product‑oriented team, contributing across the full development lifecycle and the broader platform, from requirements analysis and story refinement, through discussions with external partners and model evaluation (e.g., MMLU), to implementation and deployment.
• Design and implement on‑device AI inference pipelines for open‑source ONNX models using Microsoft ONNX Runtime GenAI (OGA) with execution providers such as AMD Ryzen AI, Qualcomm QNN, Intel OpenVINO, and NVIDIA CUDA.
• Serve as a development point of contact for E2E testing, performing hands‑on triage and root‑cause analysis of failures in on‑device AI systems, and observing first‑hand the practical limitations and trade‑offs of hybrid and on-device Conversational AI.

I was part of a team building an enterprise web platform that enabled large customers to order fully configured PCs that business professionals could use immediately. The platform supported pre‑shipment configuration of operating systems, applications, and hardware options, maintaining configuration continuity across repeated orders as hardware and software evolved. Key responsibilities:

• Worked in a product‑oriented team, contributing across the full development lifecycle, from requirements analysis and story refinement to implementation, testing, deployment, and monitoring. Responsibilities focused primarily on backend systems, with limited frontend work due to the nature of the platform.
• Served as a development point of contact for UAT and E2E testing, resolving defects and performing hands‑on triage and root‑cause analysis across services, configuration data, and test environments.
• Built a strong foundation in CI/CD and observability, leveraging GitLab pipelines, automated testing, and observability tooling to improve service reliability and surface configuration issues earlier in the delivery lifecycle.