Software Engineer, Aircall

• You have experience building production backend systems ideally in Python or Typescript (all levels welcome, 5+ years for senior roles)
• You write async code naturally, understand how to design for performance and reliability, and have shipped systems that real users depend on
• You understand what makes real-time systems different from request/response API’s. You can talk about the tradeoffs around buffering, streaming, connection management, and latency
• You've worked with LLMs or AI/ML systems in production. You understand concepts like prompt engineering, context management, function calling, and what it takes to make AI behavior reliable for customers
• You know evaluation matters and you've built or used tooling around it
• Event driven architectures, message queues, caching layers, and async task processing feel natural to you
• You've debugged distributed failures and designed systems that handle partial failures gracefully
• Testing strategy, observability, API design, and code review are things you do thoughtfully. You write code that other engineers can understand and extend
• You communicate clearly. You can explain a complex technical problem to someone without your context, write a design doc that prevents misalignment, and ask good questions before diving into implementation
• You are familiar with AI coding tools and already embedded them into your day to day engineering processes

Desirable

• At any given moment, businesses running on Aircall are fielding calls their teams can't handle alone
• Outside business hours, during peak volume, across languages, around the clock
• Our AI Voice Agent is what answers. It's not a simple IVR or a canned response bot
• It's a real time AI voice agent that understands context, talks to customers naturally, executes actions mid-conversation, and knows when to hand off to a human
• As a Software Engineer on this team, you'd work at the intersection of real time systems, production LLM applications, and voice infrastructure
• The problems are hard in a specific way: milliseconds matter when someone's on the line, latency shapes whether a conversation feels natural or robotic, and the quality of your work is immediately audible to thousands of customers
• This isn't AI prototyping. It's production AI at scale, and one of the fastest moving product areas at Aircall
• Real-time Speech Pipeline: You'll work on live audio systems: buffering, streaming, latency optimization, and the integration points with speech providers
• When latency creeps up, you investigate and fix it
• When a new provider looks promising, you evaluate and potentially integrate it
• The Conversation Intelligence: Conversation quality depends on how well we manage the LLM layer
• Prompt construction, context management, function calling, instruction ordering
• You'll build and improve the systems that make conversations feel natural: dynamic knowledge injection, action execution with branching logic, and the dialogue management that turns a phone call into something useful
• Actions and Integrations: Businesses configure AI Voice Agent to execute tasks during calls
• Look up account data, create tickets, check order status
• You'll work on the action framework: configurable API calls, success/failure branching, auth management and the runtime execution engine
• Knowledge and Memory: A voice agent is only as good as what it knows
• You'll work on how agents ingest, store, and retrieve knowledge at runtime
• There's also the question of memory: how agents retain information across a conversation, learn from prior interactions, and use context to give better answers over time
• Agent Lifecycle and Configuration: Customers need to create, configure, test, and deploy voice agents easily and intuitively
• You'll work with our designers to translate cutting edge advanced AI concepts and features into smooth, reliable and easy to use product experiences
• Evaluations and Quality: Voice AI quality is measurable
• You'll help build evaluation frameworks for each model involved, post-call analytics, call quality metrics, and Datadog instrumentation
• You'll also be part of the on-call rotation