Senior Software Engineer (Multiple Teams, Europe)

Coder

Get hot jobs first on Telegram

New positions appear faster in our channel

Location: United Kingdom
Job Type: full-time
Work Format: 🌍 Remote EU
Salary: £69K – £126K • Offers Equity • Offers Bonus • Multiple Ranges
Posted: October 15, 2026

Job Description

As a Senior Software Engineer on Coder’s Engineering team, you’ll build and evolve the systems behind our agentic development experience. You’ll work across the agent harness, integrations, and workflows that connect agents with real development environments.

You’ll stay hands-on, solve complex technical problems, and work closely with Product, Design, and other engineers to ship reliable agentic experiences.

What you’ll do here

Design and build production systems in Go, with work across React and TypeScript where needed.
Improve agent execution, tool use, context management, streaming, and long-running workflows.
Extend our provider-agnostic architecture as models and capabilities change.
Build reliable integrations between agents, workspaces, tools, and developer infrastructure.
Own projects from implementation through rollout and iteration.
Contribute to design reviews, code reviews, and technical discussions.
Partner with Product and Design to turn agent capabilities into useful developer experiences.
Improve the reliability, performance, and operability of agentic systems.

What we’re looking for

Strong experience building and operating production software systems.
Hands-on experience with Go.
Experience with React and TypeScript.
Experience building systems around LLMs or agentic workflows.
Familiarity with model APIs, tool calling, context management, or agent loops.
Good understanding of distributed systems and production reliability.
Working knowledge of AWS.
Strong problem-solving skills and comfort working through technical ambiguity.
Someone who contributes beyond their own code through reviews, collaboration, and knowledge sharing.

Bonus tacos if you have

Experience building coding agents, developer tools, or cloud development environments.
Experience with MCP, agent tools, or multi-agent systems.
Experience with remote execution, sandboxing, or isolated compute.
Experience building integrations across multiple model providers.
Experience with AWS, Kubernetes, Docker, or Linux.
Open-source contributions or experience working directly with customers.

Our tech stack

Backend: Go, Postgres
Frontend: TypeScript, React
Infrastructure: AWS, Kubernetes
Observability: Prometheus, Grafana
CI/CD: GitHub Actions

Coder’s agentic architecture is provider-agnostic. The foundation is built; this role helps extend it, improve it, and ship the next set of agentic capabilities.

About Coder

Coder is the leading platform for AI development Infrastructure, enabling enterprises to securely run human and AI-driven development workflows in consistent, governed environments. Coder provides self-hosted, agent-ready workspaces that unify developer productivity and platform governance. With Coder, enterprises can confidently evolve from human-only development to AI-assisted and autonomous workflows, without sacrificing security, compliance, or performance. Learn more at coder.com.

Interview process

We believe that the interview process should be transparent, consistent, and enjoyable. We value your time and hope to complete the interview process in two to four weeks, if schedules allow. Through your interviews, you will meet a mix of individual contributors, managers, and senior leaders.

AI use during the interview process

As an AI company, Coder embraces the use of AI tools, and we want to be transparent about our expectations as you navigate our interview process.

Not permitted: Using AI assistance during conversational interviews.

Permitted: Using AI tooling for take-home assessments. Please flag where and to what extent it was used in your take-home. Your submission will not be penalized for using AI as long as it is done honestly.

Our use of AI in hiring

We use AI tools to help manage our recruitment process efficiently and fairly. Specifically:

Ashby helps us review inbound applications by surfacing candidates who best match the role requirements we've defined. This tool does not make hiring decisions; it helps our team prioritize which applications to review first.
Granola takes notes during our interview calls so our team can focus on the conversation with you.

All hiring decisions are made by humans. Our team reviews applications, conducts interviews, and makes final selections. AI tools assist us but never replace human judgment, and these practices are conducted in compliance with applicable data protection, AI governance, and labor laws. Your data is not used to train AI models.

In accordance with New York City Local Law 144, an independent bias audit has been conducted on "Automated Employment Decision Tools"; results are available for Ashby.

If you're applying for a role at Coder and have questions about how we use AI in our process, or if you'd like to request information about the data we collect, please contact careers@coder.com.

Coder is a proud Equal Opportunity Employer

We are committed to providing equal employment opportunities to qualified applicants and do not discriminate on the basis of race, color, ancestry, religion, sex, pregnancy, gender, gender identity, gender expression, sexual orientation, national origin, age, marital status, genetic information, disability, protected veteran status or any other characteristic protected by federal, state, or local laws.

🎯 Who is this job for?

Senior-level software engineers are suitable for this role, with strong production experience in Go, React, TypeScript, distributed systems, AWS, and LLM or agentic workflows.

They should be skilled in model APIs, tool calling, context management, agent loops, reliable integrations, observability, and performance optimization.

Candidates should be familiar with owning projects end to end, building and operating agentic systems, collaborating across teams, and contributing through design reviews, code reviews, and technical discussions.

💬 Potential Interview Questions

How would you design a provider-agnostic agent harness in Go that supports multiple LLM APIs and tool-calling formats?

I would define stable internal interfaces for models, messages, tool calls, streaming events, and execution results, then implement adapters for each provider. The harness should normalize provider-specific behavior while preserving capabilities, enforce timeouts and cancellation, and expose consistent telemetry and error semantics.

How would you implement reliable context management for a long-running coding agent?

I would track token budgets, conversation state, tool outputs, and task milestones explicitly. Context would be summarized or compacted when necessary, with important constraints and files preserved separately from transient interaction history to avoid losing critical information.

How would you design streaming between an agent, workspace, and React/TypeScript frontend?

I would use a typed event protocol over a streaming transport such as Server-Sent Events or WebSockets. Events should be ordered, resumable where practical, and include identifiers for runs, tool calls, and messages so the frontend can safely handle retries, partial results, and reconnects.

What reliability mechanisms are important for agent tool execution and long-running workflows?

Each operation should support deadlines, cancellation, retries only for idempotent actions, and durable state transitions. I would also use idempotency keys, bounded concurrency, queue backpressure, and recovery logic so a worker restart does not duplicate side effects or lose progress.

How would you safely execute agent-generated commands in remote development environments?

I would isolate execution using containers or sandboxed workloads with restricted network access, filesystem boundaries, resource limits, and least-privilege credentials. Commands should be audited, time-limited, cancellable, and associated with an explicit workspace and user authorization context.

How would you model the persistence layer in Postgres for agent runs and tool calls?

I would store runs, steps, messages, tool invocations, approvals, and state transitions with stable identifiers and timestamps. The schema should support efficient retrieval of active workflows, transactional state updates, idempotent writes, and retention policies for large logs and generated artifacts.

How would you observe and troubleshoot production agentic systems using Prometheus and Grafana?

I would instrument latency, throughput, error rates, token usage, tool success rates, queue depth, retries, and workflow completion outcomes. Metrics should use bounded labels, while structured logs and traces provide run and tool-call correlation without exposing secrets or sensitive prompt content.

How would you design an agent loop that decides when to call tools and when to finish?

The loop would validate model outputs against a typed action schema, execute authorized tools, append results to context, and continue until a validated final response or explicit failure state is reached. It should enforce maximum steps, token and time budgets, detect repeated actions, and handle malformed or unsafe tool requests.

How would you integrate a new model provider without coupling the rest of the system to its API?

I would implement the provider behind a common interface covering streaming, structured output, tool calling, cancellation, and usage reporting. Provider-specific limitations would be represented as capabilities, allowing the planner or fallback logic to choose compatible behavior without scattering conditional code throughout the system.

How would you build a React and TypeScript interface for monitoring multiple concurrent agent workflows?

I would use typed client models and a normalized state store keyed by workflow and tool-call identifiers. The UI should render incremental events efficiently, preserve state across reconnects, clearly distinguish running, waiting, failed, and completed operations, and provide accessible controls for approval, cancellation, and retry.

📋 Job Summary

Coder is building the leading platform for secure, agent-ready development infrastructure, enabling enterprises to run reliable human and AI-driven workflows. As a Senior Software Engineer, you’ll build production systems for agent execution, tool use, context management, streaming, integrations, and long-running workflows, partnering closely with Product and Design. The stack includes Go, Postgres, React, TypeScript, AWS, Kubernetes, Prometheus, Grafana, and GitHub Actions. This is a fully remote role in Europe, offering £69K–£126K plus equity and bonus—a compelling opportunity to shape next-generation developer tools and agentic systems.