First call · Chase Robbins, Head of Engineering

Kubera Health
× Brian Keefe

Read this once before the call. Lead with the thesis, then use the company map and questions to have an actual conversation—not a recital.

Kubera is making healthcare contracts computable, so providers and payors can finally compare what they agreed to with what was actually paid—claim by claim.

~10-person team

$6.5M seed

NYC · 4 days onsite

TS / React / Node

The 30-second opening

I’m a product-oriented full-stack engineer at Dripos. My strongest work is taking operationally messy, high-consequence workflows—billing enforcement, realtime delivery, account security, and POS operations—and making their state, rules, and recovery paths explicit and reliable. Kubera is compelling because the domain has the same shape at a much bigger level: dense agreements and fragmented data that people need to trust enough to act on.

What Kubera actually fixes

Healthcare contracts, fee schedules, amendments, policy notices, claims, and remittances are usually disconnected.
That means an organization may know what it was paid but not what it was owed, or why the variance exists.
Kubera turns dense contract prose into structured, queryable payment logic; it then models, audits, and explains payment outcomes.
Provider-side use: underpayment detection, recovery, renewal modeling, policy-change impact, and price-transparency reporting.
Payor-side use: provider-contract modeling, pre-payment validation, and visibility into network economics.

Company context worth knowing

Seed round: $6.5M led by Upfront Ventures; founder/CEO Roja Garimella, MD, says the company has processed $3B+ in payments and that all customers have expanded. Treat those as company claims.
They are deliberately infrastructure-first, not a contingency-fee recovery shop: structured contract intelligence first; recovery/auditing on top.
Stated next bets: value-based-care contract management plus agentic payment recovery and auditing.
Chase’s own description: Kubera is about ten people processing hundreds of thousands of pages of payer contracts into structured data, with complex applications on top.

Why you map to this role

Operational reliability, not surface features

Billing standing: converted overdue invoices from a vague ops problem into explicit states, enforcement, warnings, lockout, and recovery paths.

Realtime redesign: moved large POS updates out of the socket/pub-sub data path into an ElastiCache-backed cache-key pattern, keeping events lightweight and improving failure resilience.

Messy inputs → trusted systems

POS scheduled printing: treated printing as a timed operational workflow tied to real prep behavior, rather than a fire-and-forget device action.

Cross-surface ownership: backend/API/web/POS/staff/online/shared-UI work—use this to show you can trace one business rule across the entire system.

Useful bridge: “The healthcare domain is new to me, but the engineering pattern is familiar: model the real workflow, make state and business rules explicit, preserve explainability, and design recovery paths so people can trust the system when data or integrations get messy.”

What they are screening for

A hands-on, backend-leaning full-stack engineer: TypeScript, React, Node, infrastructure, integrations, analytics/data modeling.
Someone who can parse complex healthcare data into actionable insights—and own a problem through product, architecture, and customer context.
Very high fluency with AI coding tools. Their engineering culture is explicitly AI-native: agents are connected to internal systems and sensitive data through controlled, tenant-aware access.
Early-stage builder behavior: they say “get it done,” customer obsession, honest feedback, and meaningful infrastructure over dashboards/wrappers.

Good “I noticed…” lines

“The interesting part to me is that this isn’t just document extraction. The real product is a system that can explain what a contract requires, connect that to a payment, and make the result actionable.”

“I like the infrastructure-first choice. A recovery workflow alone can become a services business; structured contract logic creates a compounding product layer.”

“Your internal AI post was unusually concrete: using AI with tenant-aware, read-only production access and actual operational data—not calling a chatbot an AI strategy.”

“The hard part seems like earning trust in the answer: versioned contracts, amendments, exceptions, claims matching, and a clear explanation when the system finds a discrepancy.”

Questions to ask Chase

Correctness: Where does the hardest correctness problem live today: extracting terms, tracking versions/amendments, matching claims, or explaining a discrepancy in a way a customer trusts?
Product boundary: How do you decide when a customer-specific contract edge case becomes a reusable product primitive rather than bespoke implementation?
First 90 days: What is the highest-leverage thing this engineer could own in the first three months?
Data + AI: What evaluation or human-review loop tells you that an AI-assisted contract interpretation is safe enough to operationalize?
Customer reality: When a customer sees an underpayment flag, what happens next today—and where does that workflow still break down?
Engineering: What does “AI-native” change about how the team plans, reviews, tests, and debugs production code, beyond using coding assistants?

Reality checks to get answered

Ask plainly, but late in the conversation: the job is explicitly hybrid four days/week in Union Square. Also get a real answer on expected weekly hours, on-call/incident ownership, and whether the stated flexibility works in practice. Do not volunteer medical detail; ask about the operating model first.

What does a normal week look like for engineers—and what makes it unusually intense?
How much time is customer-facing or implementation work versus core product/engineering?
What authority does a founding engineer have to improve architecture or say no to a brittle shortcut?

Do not overclaim

Do not claim healthcare domain depth, formal people management, or that you built Hermes from scratch.
Frame AI work as building and operating a customized Hermes/Jarvis-style assistant and workflow stack; focus on safe, useful workflows rather than autonomous-agent hype.
Keep Dripos claims bounded to the concrete features above. You have strong cross-surface delivery evidence; do not turn that into claims about HR management or payroll-tax/compliance ownership.

Sources checked today

Kubera careers · Provider product page · Payor product page · Funding note · Chase Robbins on AI adoption · live Ashby job-board API