• Wear different hats across the research-to-production lifecycle: ML Engineer, ML-Ops Engineer, Software Engineer, and Performance Engineer. • Own and evolve our research infrastructure end-to-end, from experiment orchestration and distributed training to model tracking, evaluation, and automated deployment, so researchers can move from idea to validated result quickly. • Build and scale distributed compute for research workloads (e.g. Ray-based training and data pipelines on Kubernetes/GCP), including managing GPU capacity across zones/regions and keeping experiment infrastructure reliable and cost-efficient. • Improve the speed and quality of our R&D through performance engineering: vectorizing and parallelizing simulators and training code, profiling bottlenecks, and driving large speedups. • Deeply understand the capabilities and tools offered by Phaidra’s internal platform and how to utilize them to best serve our customers. • Maintain clear and concise documentation of your research, products and actions. • Participate in making decisions for the medium-to-long-term vision impacting Research and Phaidra. • Mentor peers and delegate tasks within the team, owning the project delivery. • Act as a point of contact between Research and Production engineering teams to productionize new breakthroughs rapidly.

• Contribute to the next generation of AI-powered experiences at Autodesk • Help shape how Autodesk Assistant and our broader AI platform investments create meaningful value for customers across Autodesk products and workflows • Work closely with leaders across product, design, engineering, and data science to uncover customer needs, behaviours, and opportunities • Use customer insights to influence product strategy, roadmap decisions, and the evolution of Autodesk Assistant and agentic experiences • Evaluate the value, usability, trust, and effectiveness of Assistant capabilities and AI-powered experiences across Autodesk products • Establish cross-team connections to coordinate research efforts & evaluation approaches

• Lead end-to- end architecture for LLM-powered assistants, agents, Custom GPTs, RAG solutions, and AI-enabled applications across experience, model, retrieval, orchestration, integration, security, cloud, deployment, and observability layers. • Translate functional and non - functional use case requirements into architecture documents, diagrams, integration patterns, data flows, security models, deployment designs, and implementation guidance. • Select appropriate pattern s, including prompt engineering, RAG, agentic orchestration, workflow automation, traditional software logic, or model customization—based on quality, risk, scalability, latency, cost, and supportability. • Architect solutions using ChatGPT Enterprise, MCP-based apps, ChatGPT Skills, enterprise APIs, Databricks, vector search, governed data sources, and approved AI platforms. • Design agentic and event-driven solutions using OpenAI SDKs, LangChain / LangGraph, n8n, APIs, webhooks, and human-in-the-loop controls. • Develop targeted proofs of concept and reference implementations to validate architecture decisions, reduce delivery risk, and accelerate engineering execution. • Guide delivery teams through implementation and ensur e delivered solutions remain aligned with approved architecture, security controls, engineering standards, and operational requirements. • Conduct solution and architecture reviews; identify technical risks, platform constraints, security gaps, data-governance concerns, and operational dependencies; and document decisions, assumptions, tradeoffs, and recommendations. • Establish reusable reference architectures, solution patterns, technical standards, guardrails, and architecture decision records for enterprise GenAI adoption. • Define non - functional requirements covering privacy, security, responsible AI, performance, scalability, observability, auditability, maintainability, and total cost of ownership. • Define safeguards for prompt injection, data leakage, unauthorized retrieval, insecure tool execution, excessive agency, secrets exposure, and inappropriate model outputs, including appropriate human oversight for high-risk actions. • Establish secure identity and access patterns using OAuth, service identities, delegated authorization, role-based access control, secrets management, least privilege, and user-level auditability. • Define evaluation, regression-testing, red-teaming, tracing, monitoring, deployment, rollback, incident-management, and operational-readiness approaches for production GenAI systems. • Evaluate emerging AI platforms and technologies and contribute to the GenAI capability roadmap by identifying platform gaps, reusable services, standard integrations, and strategic architecture improvements. • Partner with product owners, business leaders, engineering, data, platform, DevOps, cybersecurity, privacy, responsible AI, and enterprise architecture teams to move solutions from concept to production. • Facilitate architecture workshops and communicate technical options, tradeoffs, dependencies, risks, and recommendations to technical teams and senior stakeholders. • Provide technical leadership across multiple GenAI initiatives, mentor engineers and solution designers, and support architecture consultations, design clinics, and technical office hours. • Produce concise architecture documents, technical specifications, implementation guidance, runbooks, and handover materials that support long-term operational sustainability.

• Turn raw, structured and unstructured enterprise information—such as policy documents, brochures, and clinical records—into accurate, high-quality context for AI applications • Focus on the quality, semantic structuring, and retrieval accuracy of data consumed by G.E.H.A’s AI solutions • Own the extraction, document chunking, vector indexing, and RAG retrieval mechanics that power G.E.H.A’s AI tools • Establish data enrichment, retrieval, and evaluation frameworks that ensure AI agents have fast, secure, and compliant access to business context