SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Tabs is an AI Operating System for Revenue, built for modern finance and accounting teams. The company combines deep revenue and accounting expertise with AI agents and applications to help finance teams automate revenue workflows end-to-end, with built-in controls, auditability, and human oversight.
In this role, you'll work on a fast-moving AI team, owning problems from initial exploration through production deployment. You'll turn messy financial data and ambiguous problems into working AI products, starting with simple baselines and adding complexity only when justified. You'll build evaluations that reflect real user outcomes, then use error analysis, ablations, and production feedback to continuously improve the system. You'll make practical tradeoffs across model quality, cost, latency, determinism, reliability, and maintainability. Close partnership with product and engineering teams is essential to building AI features that meaningfully reduce manual work for finance teams.
The team is small, so everyone has a hand in deciding what to build and making it work in the real world. The culture emphasizes shipping, learning from real usage, and iterating. You'll make assumptions explicit, follow evidence, and communicate tradeoffs clearly. The team welcomes better ideas regardless of source and collaborates closely in person five days a week.
QUALIFICATIONS:
Successful candidates typically have one of the following:
- Bachelor's degree in a relevant quantitative field plus 3+ years of relevant industry or applied research experience
- Relevant master's degree plus 1+ year of relevant industry or applied research experience
- Relevant PhD (3 years of substantive doctoral research considered equivalent to the experience above)
- Equivalent practical experience demonstrated through shipped systems, independent research, open-source work, or another nontraditional path
Required skills and experience:
- Strong statistical and machine learning fundamentals with good judgment about when to use classical ML, LLMs, agents, or hybrid approaches
- Experience shipping ML or AI systems end-to-end, from data and evaluation through production
- Comfort making progress with noisy data, weak labels, incomplete specifications, and imperfect supervision
- Experience across several of: classical ML, embeddings, retrieval and reranking, similarity search, model evaluation, LLM applications, and agentic systems
- Strong Python skills and ability to contribute to production software; TypeScript or modern web application experience is a plus
The company welcomes a range of backgrounds and encourages applications even if you don't meet every qualification, prioritizing curiosity, craft, judgment, and drive.
