Staff AI Engineer

About Goodfin Goodfin is an AI-native wealth platform focused on private markets. We're building intelligent, agentic systems that help accredited investors research, evaluate, and act on private investment opportunities with clarity and confidence. This is not an AI "feature." AI is the product. As a Staff AI Engineer, you'll be a technical leader responsible for designing, building, and hardening the core intelligence systems behind Goodfin—systems used by real investors making real financial decisions. This is a hands-on role for someone who wants to work at the boundary of what's reliable in applied AI—and make it production-grade. You'll own end-to-end systems: from architecture and modeling decisions through deployment, evaluation, and iteration. You'll also help define technical standards and shape how we build AI as a company. What You'll Work On • Architecting LLM-powered, agentic systems for private market research, analysis, and decision support • Designing hybrid reasoning pipelines that combine LLMs with retrieval, structured financial data, rules, and tools • Building robust RAG pipelines over unstructured, noisy, and proprietary data (PPMs, decks, filings, internal memos) • Developing evaluation frameworks for reasoning quality, faithfulness, latency, and cost • Implementing observability, debugging, and failure-handling for multi-step AI workflows • Partnering closely with business and design to translate ambiguous user needs into reliable intelligent behavior • Raising the bar on AI engineering practices across the team through example and mentorship **Concrete Example **Build an AI deep research analyst that can synthesize deal documents, market data, news articles, and comparable deals into actionable, well-sourced insights—while surfacing nuances rather than hiding it. Why This Is Hard • Product intuition matters: We're building a sticky, high-value product for real investors—not a demo or internal research tool. • High-stakes domain: Private market investing requires accuracy, explainability, and calibrated uncertainty. "Mostly right" is not acceptable. • Data complexity: There is no clean source of truth. Data is fragmented, sparse, and often contradictory. • Reasoning over generation: The challenge is building systems that reason, compare tradeoffs, and surface uncertainty—not just generate fluent text. • Agent reliability: Multi-step, tool-using agents must behave consistently in production, not just in demos. • Evaluation is unsolved: You'll help define what "good" looks like when traditional ML metrics fall short. • Trust as a system property: Explainability, sourcing, and failure modes are core technical requirements—not UX afterthoughts. What We're Looking For • 6+ years of software engineering experience, with deep hands-on work in applied AI / ML systems • Strong fundamentals in Python and backend system design • Proven experience with LLMs (prompting, fine-tuning, RAG, agentic workflows, or evaluation tooling) • Experience owning ambiguous, high-impact systems from concept to production • Comfort making architectural tradeoffs under real-world constraints • Ability to think at the system level while still shipping high-quality code • High product intuition and a strong sense of responsibility for user outcomes Bonus • Experience in fintech, data-intensive products, or regulated environments How We Work • Small, lean team with high ownership and minimal bureaucracy • Direct access to users and fast feedback loops • Strong bias toward clarity, correctness, and speed • High standards for technical rigor where trust matters What Success Looks Like • AI systems that customers trust and rely on—not just experiment with • Measurable improvements in reasoning quality, reliability, and latency • Clear architectural patterns that scale with product complexity • A higher technical bar across the team through example and mentorship Why Join • Work on real, unsolved AI engineering problems in production • Develop systems used for real financial decision-making • Build the technical foundation of an AI-native platform at an early but high growth stage • Competitive compensation and meaningful equity If you're excited about building serious AI systems—and want real ownership—we'd love to talk.