Remote job
AI Scientist (LATAM)
Job details
About this role
Role overview
Develop machine learning models and AI agents that generate investigative intelligence from blockchain and related data. This role covers the path from research and experimentation to production systems, seeking patterns that can help investigate illicit activity and security threats. You’ll work across intelligence categories with domain specialists, turning promising approaches into tools informed by real customer needs.
Responsibilities
- Train and assess machine learning models, including foundation models, on blockchain data. - Build AI agents and adapt language models when useful to produce intelligence at scale. - Find patterns across network, on-chain, and protocol data that reveal new connections. - Ship production-quality models, agents, and data pipelines with scalable architecture and careful evaluation. - Move ideas from prototype to production, using customer feedback to refine the work. - Manage parallel priorities and collaborate asynchronously across teams.
Requirements
- Five or more years of practical machine learning experience in applied work, research, modeling, experimentation, or evaluation. - Strong Python and SQL skills and experience shipping production machine learning code. - Experience with distributed data systems such as Spark, BigQuery, or Airflow. - A quantitative education or equivalent experience in fields such as computer science, statistics, engineering, mathematics, or physics; advanced study or research experience is a plus. - A hypothesis-driven approach to ambiguous problems, sound judgment, and ability to own work independently. - Strong technical communication in English and effectiveness in a remote, async team.
Nice to have
- Experience with language models, AI agents, or fine-tuning. - Familiarity with blockchain, cryptocurrency, fraud, money laundering, or financial crime. - Experience using AI-assisted coding tools.
Benefits and work setup
- Fully remote and globally distributed. The role requires at least four hours of overlap with Pacific business hours; most team collaboration takes place during the Pacific morning. - Work often involves fast prototypes and iteration based on customer feedback.