Applied AI Engineer

Das ist der Job

As an Applied AI Engineer, you will turn model capabilities into real product behaviour.

Darum lohnt es sich

Tech Stack Python PyTorch / JAX LLMs (OpenAI-style APIs, LLaMA, Qwen, etc.) Inference / serving (e.g., vLLM) Vector DB Ideal Experience Strong foundation in machine learning and modern neural network architectures. Collaborates effectively with engineers, product, and research teams to deliver reliable ML‑powered features.

About the Role CompanyA1 is building a proactive AI smart assistant for everyday users to bring intelligence to conversations, errands, organising and workflows. Our product focuses on achieving high reliability for long-running workflows, persistent context, and real-world task completion.

The system must handle multi-step reasoning, interact with external tools, and remain reliable despite non-deterministic model behavior. You will own problems end-to-end, from shaping model behaviour, to building the systems around it, to ensuring it performs reliably in production.

This role sits at the intersection of machine learning, systems, and product, focusing on making AI actually work for users, not just in demos, but in real-world usage. Responsibilities Design and iterate on prompts, tools, memory, and agent workflows. Turn raw model outputs into structured, reliable, and predictable behaviours.

Debug issues across the full stack (model, orchestration, infra, UX). Optimize for latency, cost, and production reliability. Develop lightweight evaluation frameworks to measure real-world performance. Work closely with product and engineering to translate ambiguous problems into working systems.

Hands‑on experience with training, fine‑tuning, or deploying ML models. Ability to write clean, production‑quality code. Comfort working across abstraction layers (model → infra → product). Strong problem‑solving skills in ambiguous, fast‑moving environments. Bias toward shipping, iteration, and continuous improvement.

Outcomes ML models in production meet expected accuracy, latency, and reliability targets. Production issues are identified quickly, debugged effectively, and root causes addressed. Data pipelines, training loops, and inference systems are robust, reproducible, and maintainable.

Iterations on models and systems are driven by real‑world signals and measurable improvements. Location Zurich #J-18808-Ljbffr