Director of Product

73 Strings•New York, NY

About The Position

You will manage some of the ley initiatives of AI product management at 73 Strings, reporting to the Chief Product Officer. You will help refine our AI product strategy, build and execute it with rest of the team, and be accountable for shipping products that materially expand what our platform can do for customers. This role demands a rare combination: someone who can set the strategic vision and then open a code editor to prototype it the same afternoon. You will own the AI product roadmap and be the person on the team who most deeply understands the technical substrate — how agents fail, where latency hides, why a workflow that works in a notebook breaks in production. You will talk to customers in the morning and be tinkering with a new orchestration pattern or evaluation framework by the afternoon.

Requirements

7+ years' experience in Product Management
Has shipped at least one production-grade agentic system — multi-step, tool-using, with real enterprise users depending on its outputs. Not a prototype. Not a demo. Something in production that had to be reliable.
Deep hands-on fluency with agentic architectures: knows the failure modes of tool-calling agents, has debugged runaway loops, has designed retry and fallback logic, and has strong opinions on when to use a single large model versus a multi-agent pipeline.
Can build and debug AI pipelines independently. Writes prompts, designs evals, reads traces, and tests hypotheses without waiting for an engineer. Comfortable enough in Python (or similar) to prototype a workflow before specifying it for the team to productionise.
Strong customer empathy and the ability to translate ambiguous, complex enterprise workflows into clear product requirements.
Strategic thinking combined with ability to translate strategy into executable initiatives.
Treats token cost as a product metric: has meaningfully reduced inference spend through prompt compression, caching, batching, or smarter model routing — without degrading quality.
Obsessed with determinism in AI systems: understands that “usually correct” is not good enough in financial workflows, and has the pattern library (structured outputs, schema enforcement, confidence scoring, human checkpoints) to make agentic behaviour reliable.
Has a strong evaluation mindset: builds test suites before shipping, tracks regressions when models change, and can tell the difference between a model improvement and a prompt overfitting to the test set.

Nice To Haves

Familiarity with the alternative investment industry or adjacent domains (asset management, fund administration, financial data).
Experience with agentic orchestration frameworks (LangChain, LlamaIndex, CrewAI, custom DAGs) and has formed clear opinions on their tradeoffs — when to use them and when to roll your own.
Background in a domain where auditability and data accuracy are non-negotiable (financial services, legal tech, healthcare, regulatory compliance).

Responsibilities

Customer engagement and research
Develop deep, first-hand understanding of how alternative asset managers work — their workflows, pain points, and decision-making processes.
Engage directly with customers and prospects to identify where agents and AI-driven capabilities can solve real problems and create measurable value.
Translate qualitative customer insight and quantitative usage data into a clear, data-driven picture of what to build and why.
AI product strategy
Define the AI product vision and roadmap: which products and features to build, in what sequence, and why.
Couple deep understanding of customer needs with a working knowledge of what is technically feasible with current and near-future AI capabilities - particularly agentic architectures, LLMs, and structured data extraction.
Make prioritisation decisions that balance short-term customer value against long-term platform differentiation.
Design, build, and ship
Work with engineering, design, data, subject matter experts, and commercial teams to take AI products from concept through to launch.
Own the end-to-end product development lifecycle: requirements, design, ultra-rapid prototyping, development, testing, launch, and iteration.
Champion a data-driven approach to product development: define success metrics, run experiments, and iterate based on evidence.
Establish best practices around AI governance, reliability, and safety in partnership with the CISO
Technical craft and AI engineering
Design and prototype agentic workflows yourself — tool-calling pipelines, multi-agent orchestration, memory and context management — not just specify them for engineers to build.
Drive token optimization across all AI features — prompt compression, semantic caching, batching, model-tier selection — treating inference cost as a first-class product constraint.
Impose determinism on non-deterministic systems: structured outputs, validation layers, confidence thresholds, fallback logic, and human-in-the-loop checkpoints that make agent behaviour predictable enough for an auditor.
Own the evaluation culture: build and maintain eval suites, regression benchmarks, and human feedback loops that make model and prompt decisions evidence-based, not intuition-based.
Leadership
Build and lead a high-performing function as the company scales.
Collaborate effectively with engineering, commercial, and executive leadership to ensure alignment and execution.
Act as an internal thought leader on applied AI for products.