ML/AI Engineer

The Role

We're looking for an ML/AI Engineer who enjoys turning real-world data into useful product solutions. You'll join our Data Science team and work across the full lifecycle: prototyping, evaluating, shipping, and operating ML and AI features across Piano's platform. This is not a single-product role. You'll move between LLM-powered content understanding, personalization and targeting, intelligent customer workflows, and the agent systems underpinning a new generation of Piano products.

You'll help build agentic products that reason over Piano's analytics, audience, and subscription data and take real actions on behalf of our customers. You'll develop new ML and AI capabilities, from LLM-based classification to classical ML models for personalization. And you'll help keep our existing production ML solutions healthy — models that serve hundreds of millions of users and are essential for our customers' businesses.

Beyond the technical scope, we're hiring for how you think. The engineers who will do their best work here are the ones who care about feeding AI systems the right information, validating what those systems produce, and optimizing for quality, cost, and latency.

What You'll Do

Design and build agent systems that power new Piano products — tool calling, multi-step orchestration, memory and context management, and the integrations that let agents act safely on customer data

Build guardrails and human-in-the-loop patterns so agents can take real actions on customer accounts

Ma intain and improve existing ML pipelines, model training workflows, and inference services to keep them stable and performant

Build and improve classical ML models behind personalization and targeting

Investigate and resolve production issues when they arise — understanding the problem by analyzing logs, model inputs and outputs, identifying root causes, and shipping enhancements that continuously improve how our ML systems perform

Collaborate with data scientists, ML/AI engineers, product managers, and other teams across the company to deliver ML/AI solutions that solve real customer problems

Deliver clean, tested, well-documented Python code and uphold good engineering practices (Git workflows, code reviews, CI/CD)

What We're Looking For

Must-have

M.Sc. in Computer Science, Mathematics, Statistics, Data Science, or a related field

3+ years of professional experience as an ML Engineer, AI Engineer, Data Scientist, or in a similar applied ML role with meaningful time building production ML or AI systems

Fluency in Python and strong software engineering fundamentals, including Git and modern collaborative development workflows

Solid understanding of core ML concepts — algorithms, evaluation, and model behavior — and the judgement to know when a classical model beats an LLM

Experience with Docker, Kubernetes , cloud platforms (AWS/GCP), CI/CD, and observability tooling (logging, metrics, monitoring)

Hands-on experience building with LLM APIs (OpenAI, Anthropic, or similar), including prompt and context engineering, structured outputs, and tool/function calling

Hands-on experience with coding agents such as Claude Code

Strong analytical and debugging skills, with a structured approach to problem-solving in unfamiliar systems

Ability to communicate clearly in English and work with product and engineering teams

Nice-to-have

Experience with agentic AI frameworks, orchestration, and tool-use patterns ( Claude Agents SDK , Pydantic AI, or similar)

Experience with MCP — writing servers, or wiring agents to internal tools and data sources

Experience with LLM observability and evaluation tooling — we use Langfuse , but experience with LangSmith , Braintrust, or similar is fine

Experience with optimizing LLM inference for cost, latency, and quality through context engineering, model selection, caching, and batching

Experience with ML pipeline tooling (Airflow or similar)

Exposure to A/B testing infrastructure for ML and AI features

Applicants must have authorization to work in this jurisdiction without sponsorship from Piano.

Nice-to-have

Experience with agentic AI frameworks, orchestration, and tool-use patterns ( Claude Agents SDK , Pydantic AI, or similar)

Experience with MCP — writing servers, or wiring agents to internal tools and data sources

Experience with LLM observability and evaluation tooling — we use Langfuse, but experience with LangSmith, Braintrust, or similar is fine

Experience with optimizing LLM inference for cost, latency, and quality through context engineering, model selection, caching, and batching

Experience with ML pipeline tooling (Airflow or similar)

Exposure to A/B testing infrastructure for ML and AI features

Applicants must have authorization to work in this jurisdiction without sponsorship from Piano.

Nice-to-have

Experience with agentic AI frameworks, orchestration, and tool-use patterns ( Claude Agents SDK , Pydantic AI, or similar)

Experience with MCP — writing servers, or wiring agents to internal tools and data sources

Experience with LLM observability and evaluation tooling — we use Langfuse, but experience with LangSmith, Braintrust, or similar is fine

Experience with optimizing LLM inference for cost, latency, and quality through context engineering, model selection, caching, and batching

Experience with ML pipeline tooling (Airflow or similar)

Exposure to A/B testing infrastructure for ML and AI features

Applicants must have authorization to work in this jurisdiction without sponsorship from Piano.

Nice-to-have

Experience with agentic AI frameworks, orchestration, and tool-use patterns ( Claude Agents SDK , Pydantic AI, or similar)

Experience with MCP — writing servers, or wiring agents to internal tools and data sources

Experience with LLM observability and evaluation tooling — we use Langfuse, but experience with LangSmith, Braintrust, or similar is fine

Experience with optimizing LLM inference for cost, latency, and quality through context engineering, model selection, caching, and batching

Experience with ML pipeline tooling (Airflow or similar)

Exposure to A/B testing infrastructure for ML and AI features

Applicants must have authorization to work in this jurisdiction without sponsorship from Piano.

Nice-to-have

Experience with agentic AI frameworks, orchestration, and tool-use patterns ( Claude Agents SDK , Pydantic AI, or similar)

Experience with MCP — writing servers, or wiring agents to internal tools and data sources

Experience with LLM observability and evaluation tooling — we use Langfuse, but experience with LangSmith, Braintrust, or similar is fine

Experience with optimizing LLM inference for cost, latency, and quality through context engineering, model selection, caching, and batching

Experience with ML pipeline tooling (Airflow or similar)

Exposure to A/B testing infrastructure for ML and AI features

Applicants must have authorization to work in this jurisdiction without sponsorship from Piano.

Nice-to-have

Experience with agentic AI frameworks, orchestration, and tool-use patterns ( Claude Agents SDK , Pydantic AI, or similar)

Experience with MCP — writing servers, or wiring agents to internal tools and data sources

Experience with LLM observability and evaluation tooling — we use Langfuse, but experience with LangSmith, Braintrust, or similar is fine

Experience with optimizing LLM inference for cost, latency, and quality through context engineering, model selection, caching, and batching

Experience with ML pipeline tooling (Airflow or similar)

Exposure to A/B testing infrastructure for ML and AI features

Applicants must have authorization to work in this jurisdiction without sponsorship from Piano.