Caffeine (caffeine.ai)

Senior Software Engineer — Machine Learning

Posted 6 Days Ago

Be an Early Applicant

Zürich, CHE

In-Office

Senior level

Artificial Intelligence • Software • App development

The Role

As a Senior Software Engineer in Machine Learning, you'll manage real-time streaming infrastructure, build job orchestration layers, and work on the database design for agentic applications.

Summary Generated by Built In

At Caffeine.ai, we are building the world's first platform that turns natural language into full-stack, modern applications that are deployed into a sovereign cloud, the Internet Computer. Our mission is to make building software as simple as a conversation — transforming ideas into live, production systems in minutes.

A core part of caffeine is the App Market: a place where applications created through conversation can be published, discovered, remixed, and extended by other users. We are building both the platform and the initial set of high-quality applications that populate it.

We are a cross-functional team of engineers and researchers building the AI that powers this new paradigm.

About the Role

As a Senior Software Engineer — Machine Learning, you will own the layer between our agentic core and everything the user sees and touches. That means multi-agent orchestration, real-time streaming pipelines, and the persistence layer that holds the state of applications that were never manually written. This is not prompt engineering — it's the industrial-grade plumbing underneath it.

What You'll Do

Own real-time streaming infrastructure: Build and operate the SSE pipeline that delivers agentic job state from backend to client — designing for latency, reliability, and graceful failure at every step.
Build the job orchestration layer: Coordinate multi-agent workflows end-to-end — dispatch, retries, state recovery, and context continuity across long-running, non-deterministic workloads.
Design schemas and persistence strategies: Own the database layer for agentic work — jobs, artifacts, agent memory, and the user's evolving application state.
Bridge agent output to product: Transform raw agent output into the structured data models the frontend and other services depend on.
Instrument the full pipeline: Measure latency, throughput, and failure surfaces — and stay close to production behaviour across every release.

Who You Are

Streaming systems experience: You've designed real-time streaming systems in production — SSE, event-driven architectures, or similar — and you know where they fail under load.
Database-as-design-surface thinker: You think about databases not just as storage but as a design surface — schema decisions, consistency guarantees, and state lifecycle are things you get opinionated about.
Agentic/LLM pipeline experience: You've worked with agentic or LLM pipelines in a backend context and understand the operational challenges of long-running, non-deterministic workloads.
Product-aware infrastructure mindset: You care about the user-facing effect of your infrastructure choices — latency, dropped events, stale state are product problems as much as engineering ones.
High autonomy: Ambiguity doesn't stall you — you scope the surface, make a call, and ship something you can measure.
Small-team energy: You're energised by small teams where your work reaches real users within days, not quarters.

Bonus Points