Member of Technical Staff - Applied ML
Applied ML engineer at Basis, an AI-agent startup for accounting, owning end-to-end projects that make agents reason, plan, and evaluate themselves: designing multi-agent architectures, managing context/memory, and building evaluation, retrieval, and document-parsing pipelines in Python, Postgres, and LLMs. Onsite in NYC 5 days/week.
- Compensation
- $175K–$350K / yr
- Location
- New York, New York, United States
- Work setup
- Not specified
- Seniority
- Not specified
- Employment
- Full time
- Experience
- 4+ years
- Category
- Software engineering
- Posted
Skills
Must have
- 4-12 years as a machine learning engineer
- Experience at a fast-paced startup (Series A-D), a tier-1 tech company, or a hedge fund
- End-to-end LLM agent applications: benchmarking, model orchestration, and evals for agent behavior and reliability
- Structured ML experimentation: framing hypotheses, building evaluation infrastructure, iterating on measurable results
- CS, physics, math, or other technical degree from a top school
- Very clear communication
- Located in the US or Canada and able to work in the New York office 5 days a week
Nice to have
- ML products and underlying models at a fast-paced company
- Deep Python and LLM/transformer expertise
- Interest in AI's impact on accounting and finance
Full description
Member of Technical Staff - Applied ML
Company: Confidential — venture-backed technology company
Location: New York, NY (Flatiron office, in person 5 days per week)
Compensation: $175,000 - $350,000 + highly competitive equity
Employment Type: Full-time
Visa Sponsorship: Visa transfers; can sponsor all types
About the Company
The client started from the belief that AI agents would become integral to knowledge work, and that accounting (structured, high-stakes and essential to every business) would be among the first domains transformed. Three years in, the client can complete a partnership tax workbook end to end, and accountants use it daily to create complex journal entries and debug reconciliations, with capabilities improving every month.
The client has raised a $140M Series B.
The Role
As an ML Engineer at the client, you will own end-to-end projects that bring intelligence into production: the systems that help its agents reason, plan and evaluate themselves. You will have full autonomy to plan projects, define success, run experiments and decide when a system is ready to ship. This is an applied role for engineers who want to operate as researchers and builders at once.
What You Will Do
- Design and iterate multi-agent architectures that automate real accounting workflows, with clear autonomy boundaries, tool usage and fallback behavior.
- Manage context and memory across agent steps; route, evaluate and optimize models under latency, cost and accuracy constraints.
- Build scalable offline and online evaluation pipelines that run hundreds of experiments automatically, with golden tasks, labeling strategies and metrics.
- Instrument the stack to catch regressions, track error taxonomies and drive closed-loop improvement.
- Architect prompt stacks, retrieval and indexing pipelines, and document parsing into structured representations agents can reason about.
- Scope projects with concise specs, build and test end to end, and communicate progress clearly within your pod.
What You Bring
- 4-12 years as a machine learning engineer
- Experience at a fast-paced startup (Series A-D), a tier-1 tech company or a hedge fund
- End-to-end LLM agent applications: benchmarking, model orchestration and evals for agent behavior and reliability
- Structured ML experimentation: framing hypotheses, building evaluation infrastructure, iterating on measurable results