← crunchyroll / Senior Product Manager, CX AI Chat
brief / art_WGZ9W25QlMU
role
model
anthropic/claude-sonnet-4.6
created
2026-05-28T17:10
Company snapshot
Crunchyroll is the world's largest anime streaming platform, serving 100M+ fans across 200+ countries with video streaming, theatrical releases, games, merchandise, and live events. It operates as an independently run joint venture between Sony Pictures Entertainment and Aniplex (Sony Music Japan), giving it significant corporate backing and global distribution muscle. In recent years Crunchyroll merged with Funimation (completed 2022) to consolidate Sony's anime assets under one brand, substantially expanding its English-dubbed library and subscriber base. The CX Technology team is actively investing in AI-driven self-service and chat automation to scale support for a rapidly growing global fanbase. Engineering reputation is not well-documented publicly; based on the JD, the team operates a modern CX tech stack including Salesforce Service Cloud and AI chat vendors such as Sierra AI.
Team stack
Based on the JD: Salesforce Service Cloud (CRM/case management), Salesforce OMS (order management), Sierra AI (likely the primary AI chat/agent platform), a Help Center CMS (likely Salesforce Knowledge or a third-party like Zendesk/Freshdesk — uncertain), CCaaS platform (vendor uncertain), WFM tooling (vendor uncertain). API integrations are explicitly called out, suggesting REST/webhook-based connectors between CX tools and internal systems. Analytics stack likely includes SQL-accessible data warehouse (BigQuery or Snowflake — uncertain) given the emphasis on data-driven roadmap decisions. GenAI layer is likely LLM-backed conversational AI via Sierra AI or a comparable vendor. Global deployment implies multi-language/locale support in the chat and help center stack.
Likely questions (10)
| area | question | why |
|---|---|---|
| domain | Walk us through how you would evaluate and select an AI chat vendor (e.g., Sierra AI vs. alternatives) for a global streaming platform with 100M users. What criteria matter most and how do you run the RFP? | JD explicitly lists Sierra AI experience and RFP management as desired qualifications; vendor selection is a core responsibility of this role. |
| system_design | Design the end-to-end architecture for an AI self-service chat flow that handles subscription billing disputes for a streaming service — from intent detection through resolution, escalation, and post-interaction CSAT capture. | JD centers on AI chat platform ownership and CSAT improvement; billing/subscription issues are the highest-volume CX contact driver for streaming services. |
| behavioral | Tell me about a time you drove a platform migration (e.g., cloud, CX tooling, or infrastructure) end-to-end. What were the biggest risks and how did you manage stakeholder alignment across engineering, ops, and legal? | JD calls out platform migrations as a required experience; candidate's Mailchimp GCP-to-AWS migration and ICE platform work are directly relevant. |
| domain | How do you measure the success of an AI chat deflection program? Walk us through the metrics you'd instrument from day one and how you'd distinguish true deflection from frustrated abandonment. | JD explicitly requires expertise in CX metrics and data-driven roadmap decisions; deflection rate vs. containment rate is a nuanced but critical distinction for this role. |
| behavioral | Describe a situation where you had to balance the needs of a third-party vendor roadmap against your internal product priorities. How did you negotiate and what was the outcome? | JD states the PM will be the main external contact for CX tech stack vendors; managing vendor roadmap dependencies is an explicit responsibility. |
| coding | You need to write acceptance criteria for an AI chat feature that must gracefully handle out-of-scope queries (e.g., a user asking about a non-Crunchyroll product). Walk us through the user story, edge cases, and how you'd define 'done' for the engineering team. | JD requires writing user stories and acceptance criteria; graceful fallback/escalation logic is a core AI chat PM skill. |
| system_design | How would you approach integrating an AI chat platform with Salesforce Service Cloud so that chat transcripts, resolved intents, and CSAT scores flow into agent dashboards and reporting in near real-time? | JD lists Salesforce Service Cloud as a key platform; API integration experience is explicitly required. |
| behavioral | Give an example of a time you used data (usage analytics, funnel analysis, or A/B testing) to make a counterintuitive product decision. What did the data show and what did you ship? | JD emphasizes 'use data to guide the product roadmap'; candidate's Intuit work with BigQuery/SQL and 275% YoY ICE growth is a strong signal here. |
| culture | Crunchyroll's values include 'Kaizen' — continuous improvement — and 'Service' — enabling joy and belonging. How do those resonate with how you've approached your work, and can you give a concrete example? | Crunchyroll explicitly surfaces its four values (Courage, Curiosity, Kaizen, Service) in the JD; culture fit questions around these are highly likely. |
| domain | How would you prioritize the AI chat roadmap when you have competing requests from Customer Service ops (agent efficiency), Engineering (tech debt), Legal (compliance/PII), and Finance (cost reduction)? Walk us through your framework. | JD lists all four of these as cross-functional partners the PM must align; prioritization across competing stakeholders is a core competency being tested. |
Talking points
- At Intuit, I owned the ICE developer platform end-to-end — growing engagements 275% YoY to 675M+ in FY23 and scaling throughput from 6K to 50K TPS via rSocket migration. That same discipline of instrumenting the right metrics, removing friction in the critical path, and aligning engineering, design, and legal is exactly what I'd apply to Crunchyroll's AI chat deflection roadmap.
- I've built AI chat and agent orchestration from scratch: at Fintellect AI I architected a RAG pipeline with multi-provider LLM fallback routing (Claude, GPT-4, Gemini) and domain-scoped conversational agents; at StreamIO I built the OpenClaw multi-agent framework with gateway protocol and subagent delegation. I understand the full stack — from intent classification and context management to escalation logic and structured output validation — not just the PM layer.
- My aeval platform (FastAPI, TimescaleDB, Redis, Ollama) gave me hands-on experience designing evaluation frameworks for AI models — factuality, safety, instruction-following, refusal detection — with statistical rigor (bootstrap CIs, Welch's t-test, Cohen's d). For a CX AI chat role, that translates directly to building the measurement infrastructure to know whether your AI is actually resolving issues or just deflecting frustrated users.
- At Splunk I delivered the Scheduler Service end-to-end in ~4 months and led a query performance initiative that achieved up to 10x improvements for a beta enterprise customer — both examples of shipping under pressure with clear acceptance criteria and cross-functional alignment. I'm comfortable writing PRDs, user stories, and acceptance criteria that engineers and non-technical CX ops stakeholders can both act on.
- I've managed third-party platform integrations and vendor relationships at scale — including the Mailchimp GCP-to-AWS migration at Intuit and API integrations across Stripe, Kinde, ElevenLabs, Redfin, and Zillow in my own products. I know how to negotiate vendor roadmap dependencies, write integration specs, and hold external partners accountable to delivery timelines.