Data Scientist

About Talio

Talio builds AI agents that help food and beverage companies sell more, innovate faster and run better. We are building the AI operating system for the industry: specialised agents that research markets, find opportunities, analyse trends, support sales, forecast operations and automate everyday work.

We are a small team in Ireland working with mid-market and enterprise customers who are moving AI agents out of pilots and into the systems that run their business. Small team. Big ambition.

About the role

Data science sits at the core of a lot of what Talio does.

Our customers want to know what is going to happen next, what is changing in their market, and what they should do about it. That can mean forecasting demand and production, optimising stock and capacity, identifying emerging trends, finding patterns across hundreds of thousands of products, or turning messy operational data into a decision someone can actually act on.

You will work across both sides of Talio. On the operations side, you will build forecasting and optimisation systems using customer ERP, production, inventory and commercial data. On the insights side, you will work with our external datasets to identify trends, opportunities and changes across products, ingredients, menus, companies and markets.

This is not a dashboard or reporting role. We want someone who can take an ambiguous business problem, work out what the data can tell us, build the right model or analytical system, and work with engineering to put it into the product.

This is a hybrid role. We work together from our office on Harcourt Street in Dublin city centre, close to the Green Luas line, with a minimum of three days per week in the office. We are an early-stage team and believe being together matters at this stage of the company.

What you’ll actually do day to day

• Build forecasting systems that customers can actually use. Work with historical orders, production, inventory, seasonality and other signals to forecast demand, stock requirements, production needs and capacity.

• Solve optimisation problems. Help customers answer questions like what should we produce, when should we produce it, how much stock should we hold, and where are we wasting capacity or resources.

• Find the signal in noisy markets. Use product, ingredient, menu, social and other market data to identify emerging trends, category shifts and opportunities before they are obvious.

• Turn data into insights, not just outputs. Build systems that explain what is changing, why it matters and what someone should consider doing next.

• Work with messy real-world data. Customer ERP and operational data is rarely perfect. You will need to understand it, clean it, join it and decide what can reliably be learned from it.

• Build models into the product. Work closely with engineering so that forecasting, optimisation, recommendation, classification and analytical systems become repeatable parts of Talio rather than one-off analyses.

• Develop better ways to rank and prioritise opportunities. Help our agents understand which trends, products, companies or operational issues are actually worth a customer’s attention.

• Evaluate whether our intelligence is good enough. Develop ways to measure forecast accuracy, confidence, trend strength, model performance and the quality of AI-generated insights.

• Work directly with customers when needed. Understand the decision they are trying to make before deciding what model or method should sit behind it.

Our team and culture

We are a small team in Ireland, and the words the team uses about working here are fast, connected, collaborative, organised, inventive and adaptable. Data science, engineering, deployment and commercial work on problems together, and nobody is far from a customer.

We care about solving the right problem more than using the most interesting technique. A simple forecasting model that changes how a factory plans production is more valuable to us than a sophisticated model nobody trusts or uses.

We hope you

• Have at least 4 years of professional data science experience, working on problems that have moved beyond notebooks and into real use.

• Have strong experience with forecasting, time series or predictive modelling. Experience with demand planning, operations, manufacturing or supply chain data is particularly valuable.

• Are comfortable with optimisation problems, including constraints, trade-offs, capacity, scheduling or resource allocation.

• Are strong in Python and modern data science tooling, and can work closely with software engineers to get models into production.

• Understand machine learning fundamentals well enough to choose the right approach rather than automatically reaching for the most complicated one.

• Are comfortable working with large, imperfect datasets and can tell the difference between a useful signal and noise.

• Think about the decision, not just the model. An accurate forecast is only useful if it helps someone decide what to do differently.

• Can communicate complicated analysis simply, including to customers and colleagues who are not data scientists.

• Are interested in modern AI systems. Experience with LLMs, embeddings, retrieval, recommendation systems or AI evaluation is useful, but not essential.

• Are curious about food and beverage, manufacturing and how these businesses actually operate.

• Have the right to work in Ireland. We are not able to provide visa sponsorship for this role.

The good, and the honest truth

The good part is that your work will be very close to real decisions. A model you build might influence what a manufacturer produces next week, how much stock they hold, where they see a capacity problem, or which emerging category an innovation team decides to investigate.

You will also have unusual access to data. Talio sits across internal operational data and large external datasets covering products, ingredients, menus, companies and market signals. There are a lot of genuinely difficult data science problems still to solve.

The honest part is that the data will not arrive clean and the questions will not always arrive well defined. Customers will sometimes ask for a forecast when the real problem is planning, or ask for an insight when the underlying data cannot support the conclusion they expect.

You will need to be comfortable challenging assumptions, working through ambiguity and choosing practical solutions over perfect ones.

If that sounds like the kind of data science you want to do, send us a short note on a forecasting, optimisation or analytical problem you have taken from messy data through to something people actually used, and why this role interests you.