Staff AI Analyst

What's the opportunity? 

Intercom’s AI Group is responsible for defining new ML features, researching appropriate algorithms and technologies, and rapidly getting first prototypes in our customers’ hands.

We are extremely product-focussed. Our team of 50+ ML scientists, ML engineers, designers and researchers works in partnership with other teams across the whole company. We move to production fast, often shipping to beta within weeks of a successful offline test. 

We constantly run experiments and measure the success of our AI features. We use frequentist and Bayesian approaches. We create dashboards to track results. We dive deep into exactly how users are being successful, and have to tease out all sorts of complex user interactions. We have to deal with inherently stochastic AI products, and complex effects.

We are looking for a dedicated analyst / data scientist to join the AI Group to help us with that.

This role focuses on the analysis and understanding of our AI features. If you want to build AI features, consider our Scientist roles.

What will I be doing? 

• Identify areas where ML can create value for our customers
• Conduct exploratory data analysis and research
• Help scientists and engineers analyse and visualize their AI changes (e.g. produce SQL queries, Jupyter notebooks, Tableau/Streamlit dashboards)
• Use cutting edge AI tools (such as LLMs) to gain leverage and understanding of our space
• Help the AI Group, and our Chief of AI, to understand core business and product questions

What skills might I need? 

• 5+ years of analytics / data science experience. Ideally, in a fast-moving environment (like Intercom), but it's also fine if you worked on interesting and complex problems in something bigger (like a bank, an insurance company, or a consulting firm)
• Strong data wrangling skills. You can connect to a data warehouse like Snowflake, and pull the data you need to analyse
• Excellent storytelling skills. When doing your analyses, you can build a compelling case using clear narratives and graphs/dashboards
• Extremely sharp applied statistical skills. You can instantly spot a prosecutor's fallacy, or consider when you might be seeing evidence of a Simpson's paradox
• Intermediate programming skills. You are comfortable with writing code (mostly) autonomously, considering edge cases and knowing how to test your work
• Solid theoretical foundation in stats and ML. We don't expect you to know how an LLM like Mistral works under the hood (although it's great if you do know it!), but you should be well-familiar with things like classification and regression, as well as testing of statistical hypotheses
• Scientific mindset. You are naturally critical and rely on data and evidence when making judgements. You are ready to abandon ideas that turned out to be bad, even if they were your own. You have a strong understanding of scientific rigor
• Strong communication skills. You mostly will be communicating to members of technical teams, but occasionally you'll also have to explain something to a less technical person

Bonus skills & attributes 

• Advanced education in math/stats, or related field (e.g. PhD or MSc)
• Strong track record of business analytics or product analytics

Benefits

• We are a well treated bunch, with awesome benefits! If there’s something important to you that’s not on this list, talk to us!
• Competitive salary and equity in a fast-growing start-up
• We serve lunch every weekday, plus a variety of snack foods and a fully stocked kitchen
• Regular compensation reviews - we reward great work!
• Pension scheme & match up to 4%
• Peace of mind with life assurance, as well as comprehensive health and dental insurance for you and your dependents
• Flexible paid time off policy
• Paid maternity leave, as well as 6 weeks paternity leave for fathers, to let you spend valuable time with your loved ones
• If you’re cycling, we’ve got you covered on the Cycle-to-Work Scheme. With secure bike storage too
• MacBooks are our standard, but we also offer Windows for certain roles when needed.