About the role

Halcyon Data is hiring a data scientist in New York, NY. This is an applied role: the output is a decision that gets made differently or a model that runs in production, not a deck. You will work close to product and be expected to explain a result to someone who will act on it.

The US market for this role is deep and well paid, and candidates rightly compare offers carefully. The band below is the real band — we do not hold a number back for whoever negotiates hardest.

What the work actually looks like

Expect experiment design and analysis, forecasting and segmentation work, and the shipping and monitoring of the models that earn their keep. A good share of the job is asking whether the question is the right one before spending three weeks answering it. The team works a hybrid week out of New York, NY, usually three days together and the rest wherever you work best.

What we look for

This is a senior individual-contributor role. You will be expected to push back on requirements that do not make sense and to set the technical direction for your part of the product. We are looking for real statistics and experiment design ahead of deep learning fluency, strong SQL, and the discipline to quantify uncertainty rather than round it away. Being able to say "we cannot conclude that" is a senior skill here.

Working at Halcyon Data

Core hours are set by the team rather than by the company, and we are deliberate about protecting long stretches of focus time. The range shown is the annual gross salary band for this role.

Responsibilities

Design, run and interpret experiments that change what the team does next
Build and ship models where a model is genuinely the right answer
Monitor deployed models for drift and degradation
Turn ambiguous business questions into ones the data can answer
Explain results and their uncertainty to non-technical stakeholders

Requirements

Applied statistics: inference, experiment design, causal reasoning
Strong SQL and confident Python with the usual scientific stack
Experience taking at least one model from notebook to production
Ability to communicate findings clearly in writing
Healthy scepticism about your own results

Skills

Python SQL Statistics A/B Testing scikit-learn Pandas

Benefits

Medical, dental and vision cover from your first day
401(k) with company match
Equity in a company with real revenue
Paid time off with a mandatory two-week minimum

HD

Halcyon Data

New York, NY501-1000 peopleData Infrastructure

Halcyon Data runs the warehouse and streaming layer for consumer subscription businesses that have outgrown a spreadsheet but do not want to hire a data platform team. We ingest events, model them into something analysts can trust, and answer questions about churn and lifetime value that used to take a quarter.

Correctness matters more than throughput here. A pipeline that is fast and quietly wrong costs a customer a board meeting, so we invest heavily in lineage, replay and the unglamorous work of handling late and duplicated records.

We are headquartered in Manhattan with engineering in Seattle and an APAC team in Sydney that owns follow-the-sun on-call.

View company profile Visit website ↗