← baseten / Product Manager, Enterprise
brief / art_FMgiUTbL4RE
Company snapshot
Baseten is an AI inference infrastructure company that enables engineering teams to deploy, serve, and scale ML models in production. The company powers mission-critical inference for high-profile AI-native companies including Cursor, Notion, Abridge, Clay, and Writer. In the last 12–24 months Baseten raised a $1.5B Series F led by Altimeter Capital, Conviction Partners, and Spark Capital, signaling aggressive growth and a move upmarket toward larger enterprise accounts. The company is known for a strong engineering culture and developer-first tooling, and is now actively building out its product function — meaning early PMs will have outsized influence on how product operates. Specific internal projects, team size, and named engineering leaders are not publicly confirmed.
Team stack
Based on the JD and public signals, the inference platform is likely built on Kubernetes-based GPU orchestration (likely NVIDIA Triton or custom serving runtimes), multi-cloud deployment across AWS, GCP, and Azure. Enterprise surface likely includes SSO/SAML/OIDC (Okta, Azure AD), RBAC, VPC/PrivateLink, and SOC 2 Type II compliance tooling. Billing infrastructure is likely usage-based (per-token or per-compute-second), possibly built on Stripe or a custom metering layer. Internal tooling likely spans Terraform/Pulumi for infra-as-code, GitHub-based GitOps, and observability stacks (Datadog, Prometheus). The product function itself is nascent — tooling choices for roadmapping and customer research are likely still being established. All stack inferences are based on the JD and industry norms for ML inference platforms at this scale.
Likely questions (10)
| area | question | why |
|---|---|---|
| behavioral | Tell me about a time you turned a stream of one-off deal-blocking requirements into a sellable, supported product capability. What was the process and what was the outcome? | The JD explicitly calls out converting 'deal-by-deal scramble' and 'one-off exceptions' into priced, supported products — this is the core job to be done. |
| domain | Walk me through how you would design an enterprise deployment options matrix for an AI inference platform — covering multi-cloud, VPC/PrivateLink, BYOC, and on-prem. How do you price and package it? | The JD lists deployment models as a primary enterprise readiness surface and asks for someone who has shipped multi-cloud and multi-region deployments. |
| domain | How have you approached SOC 2 or HIPAA compliance as a product problem rather than a legal/security checkbox? What artifacts did you own and how did you communicate posture to enterprise buyers? | The JD explicitly requires shipping SOC 2/HIPAA compliance surfaces and being credible with security and compliance buyers. |
| system_design | Design a usage-based billing and spend controls system for an enterprise AI inference platform. How do you handle metering at the token/request level, budget alerts, hard caps, and multi-team cost allocation? | The JD calls out 'billing and spend controls their finance teams expect' as a core enterprise readiness surface to own. |
| domain | Describe your experience with SSO, RBAC, and identity federation (SAML, OIDC, SCIM). How did you spec and ship these as product features, and what did the enterprise buyer conversation look like? | The JD lists 'identity and access' as a primary compliance/security surface and requires credibility with IT security buyers. |
| behavioral | Baseten's product function is nascent. How have you previously established PM norms, rituals, and standards in an engineering-led company? What did you put in place first and why? | The JD states 'you'd be one of the people who defines' the product function and that you 'set the standard for what product looks like here.' |
| coding | You're reviewing a PRD for a new VPC PrivateLink integration. An engineer pushes back saying the proposed API design will create a breaking change for existing customers. How do you evaluate the tradeoff and what do you do? | The JD requires being technical and credible with infrastructure engineers; Baseten's engineering culture means PMs must engage at the API/design level. |
| behavioral | Tell me about a developer platform or SDK you owned where you used telemetry and usage data to drive prioritization. What did you measure, what did you find, and what did you ship? | Felix's Intuit background (BigQuery, SQL, ICE telemetry across 675M engagements) is directly relevant; Baseten will want to know how he operationalizes data-driven PM. |
| culture | Baseten is customer-obsessed and moves fast. Describe a situation where customer and deal reality contradicted your roadmap assumptions. How did you handle it and what changed? | The JD calls out being 'a rigorous researcher who lets customer and deal reality drive the call' — this is a stated cultural value. |
| system_design | How would you build a compliance posture dashboard for enterprise customers — surfacing audit logs, data residency controls, encryption status, and certifications — as a self-service product experience? | The JD calls out 'compliance posture they can trust' as a core deliverable and references making readiness a 'supported, priced product instead of bespoke work.' |
Talking points
- At Intuit I owned the ICE Self-Service platform end-to-end — DevPortal, GitOps config, and onboarding tooling — and reduced developer onboarding from 2–3 weeks to minutes in pre-prod and under 24 hours for production, while mitigating $1M+ in projected opex. That's the same motion Baseten needs: turning bespoke enterprise onboarding into a supported, self-service product. I also drove 275% YoY growth in ICE engagements, scaling to 675M+ in FY23 and pushing throughput from 6K to 50K TPS via rSocket migration — so I understand what it means to build platform infrastructure that has to hold at scale.
- I've operated on both sides of the enterprise readiness problem: as a PM shipping developer infrastructure at Intuit (SDK starter kits, CI/CD integration, service language strategy presented to the CTO) and as a founder building production AI systems with real compliance and auth surfaces — Kinde OAuth 2.0, Stripe tiered billing, Electron SafeStorage for credential management, and multi-provider LLM orchestration with fallback routing. I know the difference between a feature and a sellable capability because I've had to price and package both.
- My RL Workbench and aeval platform demonstrate that I build with engineering rigor, not just product intuition. aeval includes bootstrap confidence intervals, Welch's t-test, Cohen's d effect size, CI/CD regression detection, and automated safety gates — running on FastAPI, TimescaleDB, and Redis. At Baseten, where the PM must be credible with world-class infrastructure engineers, I can engage at the architecture and API design level, not just the requirements level.
- I have direct experience in two of Baseten's bonus verticals: financial services (Fintellect AI — RAG pipeline, multi-provider LLM orchestration, real-time market analysis; Bank of America Merrill Lynch MBA associate; Topstep funded trader) and healthcare-adjacent regulated data (Kaiser Permanente SOA PM for 6+ years, Splunk Logging-as-a-Service at 1.7 TB daily volume for 200+ enterprise customers). I understand what regulated buyers scrutinize in a vendor's compliance posture and procurement process.
- At Splunk I owned three microservice backlogs (Search Service in Go, Search Catalog in PostgreSQL, SPL/SPL2) and delivered the Scheduler Service end-to-end in ~4 months, demoed at .conf19. I also built a repeatable RICE-based prioritization framework balancing internal partners, third-party developers, and Fortune 500 customers — exactly the multi-stakeholder prioritization challenge Baseten faces as it moves upmarket while keeping its developer-first DNA.