Machinify - Staff NLP Scientist
Requirements
• 6+ years of applied ML / AI / software engineering experience with a Bachelor's in CS, Math, Engineering or equivalent — or 4+ years with a Master's / PhD in a similar program. At least two production systems you owned end-to-end from ambiguous problem statement through measured impact, ideally including at least one LLM- or agent-based system. • A track record of driving vague problems to closure. You can point to initiatives where the brief was a paragraph, you scoped it, defined the metric, ran the work, and shipped a result that moved the business — not just a model or a PR. • Strong stakeholder fluency. You can sit with non-technical domain experts (clinicians, coders, ops leads, product), extract what they actually mean, translate it into a technical problem, and translate technical tradeoffs back into terms they can decide on. • Deep, hands-on agent engineering. You've designed agent loops from scratch, decided between single-agent and multi-agent topologies, engineered context (system prompts, tool surfaces, structured outputs, citation grounding), and debugged failure modes that other engineers couldn't. • Eval-first instincts. You don't ship without an eval; you don't believe a number you can't reproduce; and you've built eval harnesses that other engineers on the team now depend on. • Strong Python engineering. Clean abstractions, type discipline, async, tested code — at a level where junior and mid engineers learn from your PRs. • Hands-on experience with at least one major agent SDK — OpenAI Agents SDK, Anthropic SDK / claude-agent-sdk, LangGraph, or equivalent — with strong opinions on the tradeoffs and the scars to back them up. • Fluency with Claude Code / Codex as a power user — able to plan, execute, and debug non-trivial engineering tasks with these tools, including reading their source when needed. • Solid command of VS Code and git — branches, rebases, worktrees, conflict resolution, PR workflows. Not optional. • Strongly preferred • Experience defining and owning a metric that the business actually trusts — precision/recall against expert ground truth, dollar-weighted impact, appeal-survival rate, or equivalent — including the data pipeline behind it. • Prior work on long-context, citation-grounded systems where the model must point to evidence, not just answer. • Healthcare, legal, finance, or any other domain where "mostly right" is unacceptable and where findings get challenged by domain experts. • Experience setting technical direction for a small group of engineers (formal or informal tech lead), including reviewing designs, mentoring on agent patterns, and being accountable for an area's quality. • Familiarity with reasoning models (o-series, Claude extended thinking, Gemini thinking) and a sharp sense of when they earn their cost. • Production experience with caching, observability, and cost control on LLM workloads at scale. • Document understanding (OCR, layout-aware models, table extraction). • Vision-language models, multimodal retrieval. • Experience presenting technical results to executive or external (customer / regulator) audiences.
Responsibilities
• Drive vague business problems to closure. Sit with clinical leads, product, and ops to understand what's actually broken, where the money is, and what "good" looks like. Translate that into a concrete technical problem statement with a measurable target — and push back when the framing is wrong. • Define the metric before you build the system. Decide what you're optimizing (recall on overpayments? appeal-survival rate? cost per case? agreement with senior coders?), how it will be measured, what the baseline is, and what number constitutes shipping. Build the eval harness that produces it. No metric, no project. • Scope and sequence the work. Break an ambiguous initiative into a phased plan with explicit decision points, kill criteria, and dependencies. Decide what's in scope, what's deferred, and what's not worth doing — and communicate that crisply to non-technical stakeholders. • Set the technical direction for a problem area. Choose the agent topology, the context strategy, the model mix, the evaluation regime, the deterministic guardrails. Own the architectural call and the tradeoffs behind it. Other engineers — including senior ones — should be able to build against the foundation you set. • Raise the bar on agent engineering. Lead by example on context engineering, structured outputs, citation grounding, eval discipline, and cost/latency control. Review designs and PRs from other engineers on the team and leave the codebase and the patterns sharper than you found them. • Be the technical interface to the business. Present results to clinical, product, and executive stakeholders. Defend the methodology when findings are challenged. Know the domain well enough to argue with a senior coder about why a code is or isn't supported. • Use AI tooling like a force multiplier. A meaningful fraction of your day will be spent driving Claude Code, Codex, and similar tools to plan, scaffold, refactor, debug, and evaluate. We expect you to be dramatically faster with these tools than most engineers are without them, and to teach the rest of the team to be the same.
Benefits
• Hybrid role — we have a strong preference for in-office collaboration, with flexibility for exceptional candidates. • Top Medical / Dental / Vision offerings. • Tuition reimbursement. • Competitive salary, 401(k) with company match. • Meaningful equity. • A flexible, trusting environment where you'll be empowered to do your best work. • Compensation: Base salary for this L6 role ranges $180k–$260k+, based on level assessment, depth of experience, and skill match. Compensation also includes meaningful equity in a fast growing startup and the benefits above. • Equal Employment Opportunity at Machinify • We are committed to equal employment opportunity regardless of race, color, ancestry, religion, sex, national origin, sexual orientation, age, citizenship, marital status, disability, gender, gender identity or expression, or veteran status. We are proud to be an equal opportunity workplace. Machinify is an employment-at-will employer. We participate in E-Verify as required by applicable law. In accordance with applicable state laws, we do not inquire about salary history during the recruitment process. If you require a reasonable accommodation to complete any part of the application or recruitment process, please let our recruiters know. See our Candidate Privacy Notice at: https://www.machinify.com/candidate-privacy-notice/ • We are committed to equal employment opportunity regardless of race, color, ancestry, religion, sex, national origin, sexual orientation, age, citizenship, marital status, disability, gender, gender identity or expression, or veteran status. We are proud to be an equal opportunity workplace. Machinify is an employment at will employer. We participate in E-Verify as required by applicable law. In accordance with applicable state laws, we do not inquire about salary history during the recruitment process. If you require a reasonable accommodation to complete any part of the application or recruitment process, please let our recruiters know. See our Candidate Privacy Notice at: https://www.machinify.com/candidate-privacy-notice/
Apply in one click
Upload My Resume
Drop here or click to browse · Tap to choose · PDF, DOCX, DOC, RTF, TXT