> Markdown version of [/jobs/ext/3843982-senior-staff-software-engineer-back-end-inference](https://www.wearedevelopers.com/jobs/ext/3843982-senior-staff-software-engineer-back-end-inference). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Senior/Staff Software Engineer (Back-End), Inference

- **Company:** COHERE, LLC
- **Location:** Toronto, United States (Remote available)
- **Experience:** Expert
- **Salary:** $156,000.0
- **Contract:** Permanent contract
- **Skills:** Java (Programming Language), Application Programming Interfaces (APIs), Audit Trail, User Authentication, C++ (Programming Language), Software as a Service, Cloud Computing, Databases, Software Debugging, Distributed Systems, Fault Tolerance, Python (Programming Language), Key Management, Rust (Programming Language), Concurrency, Backend, Rate Limiting, Model Inference, Golang
- **Published:** October 20, 2026
- **Apply:** https://arc.dev/remote-jobs/j/redirect/psz9cisjcl

## About the Role

* You are a strong backend engineer with excellent fundamentals in distributed systems, databases, networking, operating systems, and security.
 * You have strong programming fundamentals and experience building production systems in languages such as Go, Python, Java, C++, Rust, or similar. You write clean, maintainable, well-tested code and make thoughtful tradeoffs between simplicity, performance, and correctness.
 * You have designed and operated production systems where correctness, availability, security, and performance mattered at significant scale.
 * You can take ambiguous, high-impact technical problems, define the right architecture, and drive them through to reliable production systems.
 * You have strong systems-design instincts and understand concepts such as concurrency, consistency, caching, queues, rate limiting, backpressure, failure domains, idempotency, and graceful degradation.
 * You think deeply about security boundaries and understand authentication, authorization, identity, cryptography, secrets, tenant isolation, and how small implementation choices can create significant security risks.
 * You are comfortable reasoning across backend services, APIs, databases, distributed systems, and the cloud infrastructure they run on.
 * You are comfortable diving into unfamiliar codebases and debugging complex problems that cross service or system boundaries.
 * You care about the details of production engineering, including instrumentation, capacity planning, migrations, backwards compatibility, incident response, and safe rollouts.
 * You are comfortable challenging existing architecture and evolving foundational components when the system's needs change.
 * You have a track record of taking technically difficult projects from an unclear starting point to systems that other engineers and customers depend on.

## Description

* Evolve and scale the backend systems that power Model Vault, including authentication and authorization, deployment management, billing, metering, and customer lifecycle.
 * Lead major architectural initiatives across identity, payments, usage systems, platform scalability, and other core product foundations.
 * Build highly reliable control-plane services for provisioning, configuring, scaling, and operating model deployments across clouds and hardware platforms.
 * Design APIs and distributed systems with strong guarantees around correctness, consistency, isolation, failure recovery, idempotency, and observability.
 * Build security and privacy into the platform architecture, including tenant isolation, least-privilege access, auditability, secrets management, encryption, and confidential computing boundaries.
 * Solve complex problems that span backend systems, cloud infrastructure, security, and AI inference, where conventional SaaS patterns are often not enough.
 * Drive projects end-to-end, from architecture and threat modeling through implementation, rollout, production operations, and iteration based on real customer workloads.