Staff Software Engineer

Accordion•Boston, MA

4h•$200,000 - $290,000•Hybrid

About The Position

This posting covers three levels: Staff Software Engineer, Senior Staff Software Engineer, and Executive Director, Software Engineering. We’re hiring the engineers who set the technical bar for how AI Lab builds — not just shipping an agentic system end-to-end, but deciding how every pod builds agentic systems: which orchestration patterns get reused, what “production-ready” means for an eval framework, and when a client problem needs a genuinely new architecture rather than another RAG pipeline. You’ll operate at the intersection of hands-on engineering and technical leadership. Expect to spend real time in the code and real time in the room with CFOs and PE operators defending the call you made. This is a highly technical role at every level in this posting — technical depth is the qualifying bar, not a management-track substitute for it. You will be expected to ship fast, defend your decisions, and continuously raise the bar on what AI-assisted delivery looks like. This is a high-velocity environment. Sprints are short, client expectations are high, and we don’t slow down between engagements. If you’ve spent time in large enterprise engineering organizations where velocity is constrained by process, this will feel different. The people who thrive here are those who find the pace energizing, not exhausting. They ship fast because they’ve internalized good judgment—not because they cut corners. They set a technical bar for others, use AI tools to multiply their own output, communicate clearly with non-technical stakeholders, and course-correct quickly when something isn’t working.

Requirements

Demonstrated technical ownership of a production agentic AI system with real users or clients — not a notebook, not a demo, not a POC that never shipped
Deep fluency in agentic architecture: multi-agent orchestration, RAG, tool use, structured output, and evaluation/observability, with real opinions on the trade-offs between approaches
Strong Python and TypeScript/React fundamentals, with the judgment to know when a framework is the right call and when it’s just glue code
A track record of translating ambiguous client asks into shipped systems on compressed timelines — ideally in a consulting, professional-services, or boutique AI-consultancy context
Genuine daily use of AI coding tools (Claude Code, Cursor, or equivalent) in your own workflow, with a clear point of view on where they help and where they don’t
Comfort setting technical direction for engineers beyond your immediate team, and defending that direction to skeptical stakeholders

Nice To Haves

Finance/PE domain experience: FP&A, GL data, ERP systems (NetSuite, SAP), or portfolio-company operating workflows
Public technical work James and the Lab leadership can review: GitHub, open-source contributions, technical writing
Experience building multi-tenant tools or platforms reused across multiple clients
Prior experience formally mentoring or growing other senior engineers

Responsibilities

Architect and ship the hardest agentic systems in the portfolio — multi-agent orchestration, RAG pipelines, and evaluation infrastructure for PE-backed portfolio companies
Set technical standards and reusable patterns (orchestration frameworks, structured-output contracts, eval methodology) that other pods adopt across client engagements
Own build-vs-buy and architecture trade-off decisions, and defend them to technical and non-technical audiences, including CFOs and PE operators
Build and evolve the observability and evaluation infrastructure that keeps production AI quality measurable firm-wide, not just pod-by-pod
Mentor and technically develop other engineers through architecture reviews, paired work on hard LLM problems, and code review
Travel to client sites as needed to scope, present, and stand behind the work
At the Senior Staff / Executive Director level: represent AI Lab’s technical point of view in cross-practice and client-executive conversations, and identify where the Lab should build new capability ahead of demand