AI Agent UX design Intern at Ambarella Inc in US Headquarters

Summary: Ambarella Inc is looking for a UX-product-focused summer intern to design and deliver a multimodal consumer-device support agent. The intern will define interaction behaviors for the agent, ensuring it performs support tasks intuitively and effectively while collaborating with engineering to meet real-time constraints. Responsibilities: - Define the Human Contract & Interaction Language - Craft the exact phrasing, tone, and cadence of every agent response (e.g., “I’ve detected low bandwidth. Would you like me to run a network diagnostic or adjust video quality?”) - Specify when the agent should ask clarifying questions, when to offer step‑by‑step guidance, and when to gracefully escalate to human support - Design error‑state messaging that is never blaming (“It looks like the camera feed is obstructed—can you clear the lens?”) and always provides a clear recovery path - Design Testable UX Behaviors - Map bounded support workflows (e.g., Wi‑Fi troubleshooting, video‑quality calibration, sensor health check) as finite state machines. Each state must have: - Trigger condition (e.g., user utterance, tool result) - Agent response (verbal/text output + optional visual cue) - Tool call (diagnostic script, settings API, etc.) - Exit criteria (success, failure, user cancel) - Ensure every behavior is scriptable, measurable, and reproducible—no “black‑box” interactions - Tool Integration & Demo Flow - Integrate the agent with Ambarella’s on‑device toolset: - Diagnostics tools (e.g., network ping, sensor status checks) - Settings APIs (e.g., adjust resolution, toggle HDR) - Guided troubleshooting scripts (e.g., “Step 1: Unplug the cable for 10 s”) - Build a demo walkthrough that showcases: - Seamless handoff between user input and tool execution - Context‑aware responses (e.g., “Because your last diagnostic showed weak signal, I’ll prioritize checking antenna placement”) - Graceful degradation when a tool fails (e.g., “I can’t access the sensor right now. Would you like to try a reboot?”) - Usability Testing & Reporting - Conduct structured usability tests with ≥ 10 internal participants. Measure: - Task success rate for each support flow - Time‑to‑completion per task (target: ≤ 45 seconds for common cases) - User confidence (post‑task survey, Likert scale) - Author a usability report that includes: - Quantitative metrics (success %, latency, error frequency) - Qualitative findings (quotes, pain points, “wow” moments) - Actionable recommendations for future agent iterations - Embedded‑First Refinement - Work with engineering to ensure all UX behaviors fit within real‑time constraints of the Ambarella dev kit: - Response latency: Agent must reply ≤ 200 ms after tool completion - Memory footprint: All dialogue scripts and state machines < 2 MB RAM - Power awareness: Agent enters low‑power “listen” mode after 30 seconds of inactivity Required Qualifications: - Pursuing B.S., M.S., or equivalent in Human‑Computer Interaction, Interaction Design, Product Management, or related field (graduating 2026‑2027) - Demonstrable UX/product experience: Coursework or projects in interaction design, conversation design, or service design - Proficiency with Figma (or similar) for flow mapping and prototyping *behaviors* (UI implementation not required) - Founder/operator experience or prior role where you defined how a “robot / agent” communicates with users is a strong plus - Strong writing skills—ability to craft concise, empathetic, and technically accurate agent dialogue - Basic scripting ability (Python, JSON, YAML) to define state machines and tool‑call interfaces Preferred Qualifications: - Experience designing multimodal interactions (voice + text + visual cues) - Familiarity with conversational design frameworks (e.g., Rasa, Dialogflow principles—implementation not required, just design mindset) - Understanding of edge‑device constraints (latency, memory, power) and how they shape UX Required Skills: Human-Computer Interaction, Interaction Design, Proficiency with Figma Important Skills: Basic scripting ability, Experience designing multimodal interactions Nice-to-Have Skills: conversational design frameworks, edge-device constraints