Head of Artificial Intelligence

Jecona; San Francisco (CA); On-site; USD 185,000 - 250,000; Full time

Job summary

A cutting-edge compliance technology firm in San Francisco is seeking a lead architect to build and optimize multi-agent systems aimed at regulatory filings. The ideal candidate will have over 5 years of experience in ML/AI, particularly in building graph reasoning systems. You will own the architecture and reliability of the product, ensuring compliance and high accuracy. The position offers a competitive salary range of $185k-$250k along with equity options. This is an opportunity to significantly impact an early-stage venture.

Qualifications

• 5+ years in ML/AI, particularly RLHF and retrieval systems.
• Experience building or scaling multi-agent systems.
• Research or applied work in RL or graph retrieval.

Responsibilities

• Architect multi-agent systems for reasoning and verification.
• Build graph-based retrieval for cross-document reasoning.
• Develop RL frameworks to optimize system accuracy.

Skills

ML/AI expertise

Building multi-agent systems

Graph reasoning

Experience with LangGraph

Research in RL

Education

Advanced degree in CS, ML, or equivalent

Job description

Overview

We are partnering with an AI-powered Compliance OS designed for public companies and regulated businesses to automate, draft, and verify regulatory filings. This is a pre-seed company where you are coming in at the beginning and setting the technical direction and building 0 -> 1.

They are looking for someone who is in a senior / staff / principal / lead role who has led or mentored, or someone in a more senior leadership role at a startup. The salary range is 185k-250k and can provide anywhere from 3-5% in equity.

What we are building:

• Multi-agent orchestration (serial + parallel) using LangGraph, MCP, and PydanticAI
• Graph-based retrieval (Graph RAG) across filings, entities, and historical disclosures
• Reinforcement learning for consistency, hallucination reduction, and reward-based accuracy
• Deep integration with ERP, Microsoft 365, and equity management systems — creating a single source of truth

The Role

We are looking for a true 0 to 1 builder. This will be your company as much as ours. You’ll own the product - architecture, algorithms, and reliability — designing multi-agent systems that reason over structured + unstructured data and deliver auditable, regulator-ready legal outputs.

This role sits at the intersection of applied research, product, and infrastructure.

You’ll

• Architect multi-agent systems for reasoning and verification in high-stakes workflows
• Build graph-based retrieval and long-context scaffolding for cross-document reasoning
• Develop RL frameworks to optimize accuracy and reduce hallucination
• Design domain-specific prompting and chain-of-thought templates for risk disclosure with domain experts
• Translate legal and financial logic into state machines and decision DAGs
• Ship fast, own quality, and make the system *work*

What we are looking for

• 5+ years in ML/AI (RLHF, retrieval-augmented systems, designing few-shots learning/high quality prompts to mitigate hallucination)
• Experience building or scaling multi-agent or graph reasoning systems
• Experience with LangGraph, PydanticAI, or custom orchestration frameworks
• Research or applied work in RL, CoT reasoning, or graph retrieval
• Bias toward speed, clarity, and ownership

Nice to Have

• Built in high-stakes, high-accuracy domains (finance, legal, healthcare)
• Background in legal or regulated workflow
• Advanced degree in CS, ML, or equivalent depth
• You would describe yourself as a character in Big Bang Theory