[Hiring] AI Engineer @BPD

Role Description

We're hiring a mid-level AI Engineer to help build and scale the AI systems behind Violet and our broader analytics platform called ThirdBase. You'll work hands-on with large language models, retrieval systems, and healthcare data—building features that turn a governed warehouse of claims, referrals, HCP engagement, and campaign performance into trustworthy, compliant, natural-language analytics. You'll own well-defined components end-to-end and grow into larger systems, with support from senior engineers and clear technical direction.

Responsibilities

• Build and improve LLM-powered analytics features—agentic workflows, natural-language querying, retrieval-augmented generation (RAG), and tool-use over structured (SQL/Snowflake) and unstructured (document) data.
• Contribute to text-to-SQL and semantic-layer systems that let non-technical users query complex healthcare datasets accurately and safely.
• Develop retrieval pipelines over document stores (vector search, hybrid keyword + semantic).
• Implement evaluation, guardrails, and hallucination checks—accuracy on domain data is non-negotiable in healthcare.
• Help enforce data-governance and compliance controls in the AI layer (derived-insights-only, no PHI/PII exposure, appropriate access controls).
• Integrate multiple tools and data sources (warehouse, campaign platforms, web, document repositories) into agent workflows.
• Deploy and maintain services on AWS, and monitor model performance, latency, and cost.
• Collaborate with product, analytics, and senior engineers to translate requirements into shipped features.

Qualifications

• 2–4 years building production software, including some hands-on experience with LLM/GenAI applications (RAG, agents, or text-to-SQL) — through work, internships, or substantial personal projects.
• Solid Python and SQL; comfort querying a cloud data warehouse (Snowflake a plus).
• Working knowledge of AWS and deploying services in the cloud.
• Substantive experience in LangGraph, CrewAI, n8n, or other agentic frameworks.
• Experience in AI-assisted coding and CICD processes.
• Familiarity working with Node, React, or other JavaScript frameworks.
• Exposure to retrieval systems, embeddings, or vector search.
• Experiments-driven design using evaluation harnesses for change management.
• Understanding of prompt engineering and LLM guardrails— a sense of how to make model outputs reliable, not just functional.
• Awareness of data privacy and compliance basics and willingness to build to them.
• Ability to own a feature or component end-to-end and collaborate across a team.

Requirements

• Any experience with healthcare/life-sciences data (claims, referrals, HCP, ICD-10) or another regulated data domain.
• Familiarity with HIPAA / PHI-PII constraints or de-identified / derived-insights data models.
• Exposure to agent orchestration frameworks and multi-tool workflows.
• Basic MLOps/LLMOps: monitoring, cost/latency optimization.
• Background in analytics, BI, or data engineering.

Core Tech Stack

• Cloud: AWS (e.g., Lambda, S3, ECS/Fargate, API Gateway)
• Data warehouse: Snowflake (SQL, semantic layers)
• Languages: Python (primary), SQL
• AI/ML: LLMs (OpenAI/Anthropic-class models, incl. Amazon Bedrock), RAG, embeddings/vector search, prompt engineering, function-calling / tool-use, skills
• Data domains: healthcare claims, physician referrals, ICD-10 diagnosis, HCP engagement, campaign/media performance, survey/primary research
• Data & document tooling: vector databases, hybrid search, integrations with platforms like Egnyte and Google Drive
• Analytics libraries: pandas and modern data-analysis/plotting tooling
• Governance: compliance-aware data access, PII/PHI handling, source citation