Job Description
We are building an AI-driven platform that generates defensible, policy-aligned proxy voting recommendations for institutional investors. The system combines deterministic orchestration with probabilistic reasoning, using a modular, LLM-agnostic pipeline that emphasizes correctness, traceability, and auditability.
This is a foundational hire on a very small team tackling a large, messy, high-stakes problem. We are looking for someone who takes full ownership of outcomes, thrives in uncertainty, and is willing to do what it takes to get the job done — whether that means refining prompts, debugging data issues, chasing down edge cases, or improving tooling that didn’t exist yesterday.
Core Responsibilities
• Take end-to-end ownership of the AI recommendation generation engine, including:
• Prompt design, iteration, and versioning
• Tool selection and orchestration
• Planner and sub-planner behavior
• Auditor and verification logic
• Optimize interactions between deterministic pipeline stages and probabilistic model outputs, doing whatever is necessary to improve correctness and reliability.
• Design, maintain, and evolve golden datasets with expected recommendation outputs and rationales.
• Build and run regression testing frameworks that surface meaningful changes, not noise.
• Dive into messy, real-world data to identify gaps, inconsistencies, and edge cases in proxy documents and regulatory filings.
• Actively hunt down failure modes in the system — not just when something breaks, but before it reaches users.
• Improve data extraction quality, tool usage, and prompt behavior to ensure recommendations align with internal policy logic.
• Implement guardrails, confidence scoring, abstention logic, and fallback behavior when signal quality is insufficient.
• Partner closely with the CTO to evolve system architecture while maintaining determinism, auditability, and velocity.
What This Looks Like Day-to-Day
• One day you may be deep in prompt tuning and model comparisons.
• The next day you may be debugging a subtle data edge case or rewriting evaluation logic.
• You will regularly cross boundaries between ML, data engineering, and backend systems to move the product forward.
• You will be expected to take initiative, identify what needs fixing, and fix it — without waiting for perfect specs.
What Success Looks Like
Within 3–6 months:
• The recommendation pipeline reliably processes complex proxy documents with clear visibility into remaining edge cases.
• Regression tests catch real regressions and help explain why behavior changed.
• The system produces recommendations that are consistent with historical votes and internal policy logic.
• When outputs deviate, the reasons are understandable, measurable, and actionable.
• The system is measurably more robust because you owned the hard, messy parts others tend to avoid.
Required Experience
• 5–10+ years of experience in applied ML, decision systems, or data-driven product engineering.
• Proven ability to work effectively in early-stage, low-structure environments.
• Strong experience deploying LLM-powered systems in production, especially prompt-first or hybrid architectures.
• Comfort taking ownership of ambiguous problems and driving them to resolution.
• Ability to write clean, production-quality code and iterate quickly with review and feedback.
• Willingness to roll up your sleeves and contribute wherever needed to move the system forward.
Strongly Preferred
• Background in quantitative systems, applied ML, or decision-support platforms.
• Experience in fintech, regtech, governance, or other highly scrutinized domains.
• Startup experience where “that’s not my job” was not an acceptable answer.
• Strong instincts around failure modes, edge cases, and system robustness.
Why This Role Matters
This role is critical to the credibility of the platform. You will help define how AI-driven recommendations are generated, tested, and trusted. The decisions you influence matter, and the system you help build will shape how institutional votes are made
This is a foundational hire on a very small team tackling a large, messy, high-stakes problem. We are looking for someone who takes full ownership of outcomes, thrives in uncertainty, and is willing to do what it takes to get the job done — whether that means refining prompts, debugging data issues, chasing down edge cases, or improving tooling that didn’t exist yesterday.
Core Responsibilities
• Take end-to-end ownership of the AI recommendation generation engine, including:
• Prompt design, iteration, and versioning
• Tool selection and orchestration
• Planner and sub-planner behavior
• Auditor and verification logic
• Optimize interactions between deterministic pipeline stages and probabilistic model outputs, doing whatever is necessary to improve correctness and reliability.
• Design, maintain, and evolve golden datasets with expected recommendation outputs and rationales.
• Build and run regression testing frameworks that surface meaningful changes, not noise.
• Dive into messy, real-world data to identify gaps, inconsistencies, and edge cases in proxy documents and regulatory filings.
• Actively hunt down failure modes in the system — not just when something breaks, but before it reaches users.
• Improve data extraction quality, tool usage, and prompt behavior to ensure recommendations align with internal policy logic.
• Implement guardrails, confidence scoring, abstention logic, and fallback behavior when signal quality is insufficient.
• Partner closely with the CTO to evolve system architecture while maintaining determinism, auditability, and velocity.
What This Looks Like Day-to-Day
• One day you may be deep in prompt tuning and model comparisons.
• The next day you may be debugging a subtle data edge case or rewriting evaluation logic.
• You will regularly cross boundaries between ML, data engineering, and backend systems to move the product forward.
• You will be expected to take initiative, identify what needs fixing, and fix it — without waiting for perfect specs.
What Success Looks Like
Within 3–6 months:
• The recommendation pipeline reliably processes complex proxy documents with clear visibility into remaining edge cases.
• Regression tests catch real regressions and help explain why behavior changed.
• The system produces recommendations that are consistent with historical votes and internal policy logic.
• When outputs deviate, the reasons are understandable, measurable, and actionable.
• The system is measurably more robust because you owned the hard, messy parts others tend to avoid.
Required Experience
• 5–10+ years of experience in applied ML, decision systems, or data-driven product engineering.
• Proven ability to work effectively in early-stage, low-structure environments.
• Strong experience deploying LLM-powered systems in production, especially prompt-first or hybrid architectures.
• Comfort taking ownership of ambiguous problems and driving them to resolution.
• Ability to write clean, production-quality code and iterate quickly with review and feedback.
• Willingness to roll up your sleeves and contribute wherever needed to move the system forward.
Strongly Preferred
• Background in quantitative systems, applied ML, or decision-support platforms.
• Experience in fintech, regtech, governance, or other highly scrutinized domains.
• Startup experience where “that’s not my job” was not an acceptable answer.
• Strong instincts around failure modes, edge cases, and system robustness.
Why This Role Matters
This role is critical to the credibility of the platform. You will help define how AI-driven recommendations are generated, tested, and trusted. The decisions you influence matter, and the system you help build will shape how institutional votes are made
💡 Quick Summary
Seeking a career-building opportunity? The Senior Applied ML / Decisions Systems Engineer ($200K + Equity) position is now open for candidates interested in the IT Engineer & Developer Jobs sector. This role in New York City offers a professional environment and growth potential.
Requirement Snapshot: Candidates should possess basic communication skills, a proactive attitude, and the ability to work in a team. Experience in IT Engineer & Developer Jobs is a plus.
