Director, Agent Intelligence - HYBRID ROLE

Scorpion Therapeutics; Boston (MA); Hybrid; USD 216,000 - 325,000; Full time

Benefits offered by this job

Annual bonus

Annual equity awards

Job summary

Scorpion Therapeutics is seeking a senior AI/ML data engineering leader to design and operate enterprise AI agents in a hybrid Boston environment, with the office required three days per week.

You will own memory layers, “personas as code,” tooling catalogs, and governance across prompts, tools, and responses, while partnering with executives to raise reliability, scalability, and safety of AI systems.

Qualifications

• 10+ years in data engineering, AI/ML, or analytics.
• 3+ years in NLP or AI agent/conversational AI.
• Production AI agent experience with LangChain or similar.
• Experience with Snowflake Cortex and/or Databricks.
• Full-stack AI engineering across prep, prompting, orchestration, eval/testing, deployment.
• Leadership and communication with executives.

Responsibilities

• Design and operate agent memory layer (personal/org memory) including retention, retrieval, conflict resolution, and classification.
• Define and steward “personas as code” with versioning, review, composition, and deployment.
• Curate a skills catalog of atomic tools and higher-order compositions (contracts, registry, quality standards).
• Design, build, and deploy flagship enterprise agents (Snowflake Cortex AI, Databricks AI, agent orchestration frameworks).
• Define and maintain enterprise standards for orchestration, prompt libraries, persona definitions, tool-use patterns, and response quality.
• Set conversational UX quality bar (tone/persona, citations, error handling, escalation paths, transparency) and drive consistency.
• Govern reusable prompt and skill assets with documentation and contribution workflows.

Skills

Data engineering

AI/ML

NLP

Agent development

LangChain

Snowflake Cortex

Databricks

Agent memory design

Production ops

Executive leadership

Tools

LangGraph

Job description

Hybrid (Boston office required 3 days/week).

Responsibilities:

• Design and operate Vertex’s agent memory layer (personal/org memory), including retention, retrieval, conflict resolution, and classification.
• Define and steward “personas as code” (voice, remit, boundaries) with versioning, review, composition, and deployment.
• Curate a skills catalog of atomic tools and higher-order compositions (contracts, registry, quality standards).
• Design, build, and deploy flagship enterprise agents (Snowflake Cortex AI, Databricks AI, agent orchestration frameworks) to harden primitives.
• Define and maintain enterprise standards for orchestration, prompt libraries, persona definitions, tool-use patterns, and response quality.
• Set conversational UX quality bar (tone/persona, citations, error handling, escalation paths, transparency) and drive consistency.
• Govern reusable prompt and skill assets (system prompts, templates, tool instructions, conversation patterns) with documentation and contribution workflows.

Qualifications / Required skills:

• 10+ years in data engineering/AI-ML/advanced analytics; 3+ years in NLP, AI agent development, or conversational AI.
• Production AI agent experience with LangChain/LangGraph or similar.
• Strong enterprise experience with Snowflake (Cortex) and/or Databricks (Agent Bricks/AI Functions/Mosaic AI).
• Full-stack AI engineering across prep, prompting, orchestration, eval/testing, deployment, monitoring.
• Direct experience designing agent memory/state and retrieval.
• Proven production operations (availability, monitoring, continuous improvement).
• Leadership/communication with executives.

Preferred:

• Pharma/life sciences domain experience.
• Reusable agent abstraction/platform primitive design.
• Agent evaluation tooling (Braintrust, LangSmith, Mosaic AI Eval, Arize, Ragas).
• Model fine-tuning/distillation/post-training for domain adaptation.

Compensation:

• Pay range: $216,400–$324,600; eligible for annual bonus and annual equity awards.