Coris (YC S22) Is Hiring Back End Engineer

🧠 About Coris

Coris is the AI phase for hazard and financial operations. Platforms, marketplaces, fee providers, and banks - GoFundMe, Clio, Cherry, and a expanding catalog - use Coris to onboard merchants faster, detect them continuously, and act on fraud without adding headcount. One client described us as "Cursor + Lovable for hazard teams."

We're backed by Y Combinator, Lux Capital, and Pathlight, founded by payments-domain experts, and expanding quickly on client pull. We're a small team; this employ owns a ample portion of the backend.

👥 Who you'll activity with

Coris was started by Vinodh (CEO) and Shyam (CTO). We've all spent 20+ years in fraud and hazard management, operating hazard engineering, operations and merchandise teams at Google, eBay, PayPal and WePay. You'll activity immediately alongside the two of us in Palo Alto.

🚀 Why this function matters

Every Coris decision runs on data we drag from fee processors and platforms: Stripe, Adyen, Fiserv, TSYS, Repay, and more, complete APIs, webhooks, and SFTP document feeds - hundreds of millions of events a month, billions of rows in Postgres. On top of that sits a signals and rules engine, an document graph, and the partner-facing API our customers build on.

Here's the honest type of the work. Files attain out of command complete SFTP. One processor sends dates in UTC, another in the merchant's local time, and neither says which. A processor's decrease code lives in a distinct pillar than the spec says. A webhook ships malformed JSON at 2am. A backfill employee runs out of recollection on a 40B-row table. A customer's hazard guide says "our numbers appearance off" and needs an answer today.

We're hiring the engineer who volition own all of that end-to-end, and who thinks it's the engaging part - since all integration we create reliably is one a competitor hasn't, and that's the moat. Getting the boring activity correct all period is how enduring companies are built!

🥷 What you'll do

• Own processor and phase integrations. Build and run ingestion from Stripe, Adyen, Fiserv, TSYS, Repay, and new processors as we sign them: APIs, webhooks, and lot document feeds. Make them correct, idempotent, observable, and PCI-safe.
• Own the partner API and customer-specific work. Extend the v2 API, build per-customer configuration and signals, and activity immediately alongside customers' hazard and ops teams to rotate a vague study into a fix, fast.
• Build the hazard engine. Add signals and rules to the DAG motor that scores merchants and transactions; scheme Postgres schemas, indexes, partitions, and aggregations that remain accelerated at billions of rows.
• Run it in prod. Celery workers, backfills, deadlocks, circuit breakers, latency, Sentry monitoring. You container it, you own it.
• Use AI everyplace it earns its place. Coding agents for archetypal drafts and refactors, AI assessment on all PR, and LLMs inner the merchandise (dispute agents, document parsing, organized extraction). You'll additionally assistance set the team's standards for how we use these tools.
• ~10–20% product-facing TypeScript/Next.js whenever a characteristic needs it.

📅 Your archetypal 90 days

• Week 1: your archetypal PR is merged and operating in production.
• 30 days: you're picking up significant features and shipping them.
• 60 days: you're taking on harder problems and construction genuine degree in one or additional parts of the system, akin processor integrations, the hazard motor or the partner API.
• 90 days: you're the go-to individual for a important part of the platform.

✅ You may be a fit if

Nobody matches all row below. If most of it appears akin you and the activity complete is the benevolent you desire to do, we'd akin to comprehend from you.

• A few years shipping manufacturing Python backends (Django/Postgres or close; 3+ years is a guide, not a cutoff). Ideally you've owned a scheme in manufacturing through growth: schema changes on ample tables, concurrency and locking bugs, backfills, on-call. You're at residence in SQL and can peruse a query plan, or desire to get extremely fine at it.
• An AI-native way of working. You attain for coding agents (Claude Code, Codex, Cursor or similar) by default, and you've built item alongside LLMs that group use, at activity or on your own. If you're before on this curve but moving fast, inform us what you've built.
• Judgment concerning what AI produces. You have a awareness for which changes to let an delegate run alongside (test scaffolding, a well-specified refactor, a immigration you can diff) and which to peruse row by row (anything touching money, cardstock data or PII, a schema alter on a huge table, locking and concurrency). You desire to comprehend all row of a diff before it ships, and you treat AI assessment comments as input, not instruction.
• You akin the schlep. Integrations, escalations, backfills, document formats from 1995. You see the form in the disorder and don't scorch out on it, since you comprehend why it's the product.
• Comfortable alongside customers. You'd be at ease, or eager to get there, on a call alongside a customer's hazard team, saying "give me an hour," and coming rear alongside the answer.
• Committed to the startup outcome. You desire equity that matters and you activity akin an owner: comfortable alongside ambiguity, alongside lengthy weeks whenever it counts, and alongside being in the area 5 days a week.

🙏 Nice to have

• Payments, merchant risk, underwriting, or fintech backdrop (Stripe, Adyen, Plaid, Ramp, a PayFac or ISO).
• Prior early-stage startup experience.
• Built a rules engine, indication pipeline, entity-linking / chart system, or characteristic store.
• SFTP and batch-file integrations, fixed-width or bequest formats, PCI-scoped data handling.
• Shipped LLM features in manufacturing - organized extraction, agents alongside tools, evals.

🤝 How we work

• Bias toward act and measurable impact; we container fast, measure, and iterate alongside genuine client feedback.
• Everyone owns their code in prod.
• AI-first development: agents write, humans own. Every PR gets an AI assessment and a individual one if necessary.
• In-person, 5 days a week in Palo Alto. Like any high-growth startup, we go fine beyond the customary 40–50 hours a week.