Senior Data Scientist + Machine Learning Engineer

Enterprises waste millions on accounting firms to calculate R&D tax credits and capitalize software costs. Neo.Tax is automating this entirely. Our software ingests data from project management, identity management, payroll systems, and financial systems and uses ML/LLMs to do in hours what used to take months of manual work.

Neo.Tax is seeking a Senior Data Scientist + Machine Learning Engineer (combo role) to build and ship models and production ML systems that power our core product experiences and automate complex tax and accounting workflows. This role is hands-on and product-oriented: you will take ambiguous problems, turn them into measurable objectives, build robust solutions, and collaborate closely with engineering and product to deploy and iterate in production.

We are a remote company, but we prefer to hire in time zones that can overlap with our HQ in San Francisco, CA!

• Own ML/AI problem spaces end-to-end: Define success metrics, create baselines, iterate on approaches, and drive projects from prototype to production.

• Model development: Build and improve models spanning classification, information extraction, entity resolution, clustering, ranking, anomaly detection, and forecasting.

• LLM systems: Design and evaluate prompt + retrieval + tool-calling pipelines; improve quality through datasets, labeling, and systematic evaluation.

• Data foundations: Define datasets, labeling strategies, and data quality checks; build features that generalize across customer contexts.

• Experimentation and evaluation: Design offline evaluations and online experiments; build dashboards and monitoring to detect regressions.

• Production ML engineering: Build and operate training/inference pipelines (batch and/or online), model serving, feature/data pipelines, and monitoring/alerting for quality, latency, and cost.

• Partner with engineering: Collaborate on productionization, scalability, reliability, latency, and cost; contribute directly to model-serving or batch pipelines as needed.

• Cross-functional collaboration: Work with product, engineering, and customer-facing teams to understand workflows and translate real customer pain into ML deliverables.

• Technical communication: Write clear specs and postmortems, document trade-offs, and communicate progress, risks, and decisions.

• MS/PhD in Computer Science, Statistics, Mathematics, or a related quantitative field, or equivalent practical experience.

• 6+ years of industry experience as a Data Scientist / Applied Scientist / ML Engineer shipping ML to production (or equivalent).

• Strong proficiency in Python and the modern data/ML ecosystem (NumPy/Pandas, scikit-learn, PyTorch or TensorFlow).

• Strong understanding of statistical modeling, experimentation, and evaluation (metrics, confidence intervals, A/B testing, bias/variance, error analysis).

• Experience building data pipelines and working with SQL and relational databases.

• Experience deploying and maintaining models in production (batch or real-time), including monitoring and iteration; comfortable owning operational concerns (reliability, latency, cost).

• Ability to operate with high ownership in ambiguous environments; strong communication and collaboration skills.

• Ability to effectively design and implement solutions without the help of AI (more info on how we use AI at Neo.Tax below).

• Experience with LLM evaluation, synthetic data generation, RAG, or tool-augmented agents.

• Bonus

  • Experience with information extraction and document understanding.

  • Experience with distributed data processing (e.g., Spark, Beam) and/or workflow engines.

  • Experience with GCP, AWS, or Azure.

  • Experience working at early-stage, venture-backed startups.

• Salary range: $190,000-210,000

• Stock Option Plan (Equity)

• Health Care Plans (Medical, Dental, Vision, Short-term Disability)

  • 90% coverage for individual + family

• Health & Wellness subsidy

• Retirement Plan (401k)

• Paid Time Off (Vacation, Sick & Public Holidays)

• Family Leave (Maternity, Paternity)

• Work From Home option

Still interested? Read on for more information!

• Series B preparation underway: You'd be joining at a pivotal stage where early employees have meaningful impact on the company's trajectory.

• Real traction: Multiple profitable months and 4x revenue growth year-over-year. Our Q1 was the most successful quarter in the company’s history! This isn't a speculative bet.

• Big Customers: Adobe, Brex, CapitalOne, Mercury, Notion, Thomson Reuters, and Whoop, to name a few.

• Small team, big ownership: You will own meaningful parts of the ML system that directly impact customers and revenue.

• Greenfield problems: Model diverse customer data, scale pipelines, and automate an industry that's barely been touched by software.

• Ownership-oriented: You want autonomy and responsibility. You're not looking for someone to hand you a detailed spec and check your work.

• Proactive communicator: You identify and raise risks early, summarize what you've heard, and ask clarifying questions rather than making assumptions.

• Pragmatic over idealistic: You evaluate solutions based on trade-offs, not dogma.

• Product-minded: You care about shipping improvements that move customer outcomes, not just training models.

• Comfortable with ambiguity: You can dive into unfamiliar data and systems and figure out what needs to happen.

• The data science + ML engineering team consists of four full-time team members (including you) and one part-time employee. You’ll work closely together and collaborate daily with product and engineering to ship new features.

• We're early adopters of AI tooling. Everyone on the team uses Claude Code or OpenAI Codex daily (including in sales and GTM), and we actively experiment with new AI workflows. We're looking for someone who sees AI as a force multiplier for skilled practitioners, not a replacement for fundamentals.

• Typical day

  • 9:00am — You start work. Check your email and Slack. Review notes from yesterday where you left off.

  • 9:15am — You post an asynchronous update in Slack discussing the Linear tickets you worked on yesterday and what you plan to focus on today. You highlight progress and blockers. You raise a risk related to the estimated timeline of your current project. You at-mention your counterpart on the Engineering team to notify them of a change you’re planning to make to a particular model in the prediction-service and ask if they’d like to discuss further.

  • 9:45am — Join weekly Data Science Tea Time to discuss technical topics relevant to the whole team.

  • 10:45am — Review two pull requests from your teammates.

  • 12:00pm — Lunch break.

  • 1:00PM — Meet with Firas, the CTO who’s also your manager, for your bi-weekly 1:1. Ask some questions about the quarterly goals and new enterprise customers we’re onboarding. Provide him with feedback on a new evaluation methodology we just implemented for a class of AI models in production.

  • 1:30pm — Work on your assigned project. Start a Slack huddle after 30 minutes of investigation to discuss some unfamiliar code with another data scientist.

  • 5:00pm — Take some notes on where you should pick up tomorrow. End the day.

  • If you’re starting a new project soon, you’d participate in a kick-off meeting with all the appropriate stakeholders to ensure you understand the requirements and that the project spec has all the necessary information before you start implementing so that you can deliver what’s expected.

• You understand the fundamentals of our technology stack and how ML fits into the product end to end.

• You understand our business domain (R&D tax credits and software capitalization), customer workflows, and what “good” looks like.

• You’ve shipped multiple measurable quality improvements to production models/pipelines, with monitoring in place.

• You’ve established or improved evaluation infrastructure (datasets, metrics, dashboards) that makes iteration faster.

• People who prefer research-only roles and do not want to ship production systems.

• People who avoid ambiguity and require complete, detailed specs before making progress.

• People who cannot evaluate solutions based on trade-offs and real-world constraints.

• People who rely on AI as a substitute for core statistical/ML fundamentals.