Data & AI Engineer

A P Ventures LLC; Columbia (MD); Hybrid; USD 120,000 - 170,000; Full time

Job summary

A P Ventures LLC in Columbia, MD is seeking a data/AI engineer to join a federal proof-of-concept team. You will implement protocol data structures, build ETL/ELT pipelines, and support LLM-driven information extraction and validation within strict provenance and regulatory frameworks.

You will work with SQL, JSON/XML structures, and platforms like PostgreSQL, Elsa/HALO, and Databricks to ground outputs in approved sources and demonstrate POC viability.

Qualifications

• 3–5 years in data engineering, software dev, data science, AI/ML, analytics, or related technical role.
• Hands-on experience with Python for data processing, automation, API integration, or AI/ML workflows.
• Data modeling experience and working knowledge of relational databases, XML, and JSON structures.
• Practical exposure to generative AI and large language models, including prompt development, LLM APIs, structured outputs, evaluation.
• Must be able to obtain and maintain a High-Risk Public Trust (or equivalent federal client access).
• Must be authorized to work in the United States without employer sponsorship now or in the future.

Responsibilities

• Implement and maintain protocol data structures and database components for extraction, normalization, storage, retrieval, comparison, and reviewer-facing usage of ICH M11-aligned and historical protocols.
• Convert approved standards and models into practical implementation artifacts, including schemas and mappings for ICH M11, CDISC, USDM, ontology, metadata, and terminology models.
• Build JSON structures and data-processing components aligned to guidance from senior standards and architecture SMEs.
• Develop and support ETL/ELT pipelines, Python-based transformations, and protocol content preparation workflows while preserving source provenance, metadata, validation status, and reviewer feedback.
• Support LLM workflow development and testing for protocol information extraction, summarization, classification, normalization, and comparison, including prompt configuration and structured-output generation.
• Assist with retrieval-augmented generation, embeddings, semantic search, vector indexing, evidence retrieval, and context assembly to ground responses in protocol content and approved reference sources.
• Develop SQL and Python data access logic and assist with APIs, JSON/XML processing, data validation, and automation supporting extraction, historical retrieval, comparison, and source traceability.
• Execute data- and AI-focused testing, including checks for completeness, mapping consistency, retrieval quality, unsupported or hallucinated content, ambiguity, reproducibility, prompt behavior, and regression across approved configurations.
• Document data models, prompts, mappings, test results, interfaces, transformations, implementation decisions, and technical limitations for client review, knowledge transfer, and iterative POC refinement.
• Work with senior data engineers, clinical protocol SMEs, data standards specialists, architects, and AI/prompt engineers to translate approved requirements into working POC components.
• Participate in prompt and LLM testing, technical working sessions, POC demonstrations, validation activities, and iterative refinement based on client reviewer feedback.
• Support implementation within provided platforms and approved data environments; this is a proof of concept effort and does not require independent ownership of a production platform.
• Escalate ambiguous clinical, semantic, or AI-generated results to designated SMEs instead of making independent scientific or regulatory interpretations.
• Build practical data and AI engineering skills while increasing expertise in clinical protocols, USDM/M11 standards, semantic models, and regulated AI validation.

Skills

Python

LLM APIs

Prompt development

JSON

XML

SQL

Data modeling

ETL/ELT

Education

Bachelor's degree in computer science, data science, computer engineering, or related field

Tools

Databricks

PostgreSQL

Elsa/HALO

HALO

AWS

REST APIs

Job description

Join a federal proof of concept team delivering hands-on data engineering and applied AI/LLM support for AI-assisted clinical trial protocol review.

Responsibilities

• Implement and maintain protocol data structures and database components for extraction, normalization, storage, retrieval, comparison, and reviewer-facing usage of ICH M11-aligned and historical protocols
• Convert approved standards and models into practical implementation artifacts, including schemas and mappings for ICH M11, CDISC, USDM, ontology, metadata, and terminology models
• Build JSON structures and data-processing components aligned to guidance from senior standards and architecture SMEs
• Develop and support ETL/ELT pipelines, Python-based transformations, and protocol content preparation workflows while preserving source provenance, metadata, validation status, and reviewer feedback
• Support LLM workflow development and testing for protocol information extraction, summarization, classification, normalization, and comparison, including prompt configuration and structured-output generation
• Assist with retrieval-augmented generation (RAG), embeddings, semantic search, vector indexing, evidence retrieval, and context assembly to ground responses in protocol content and approved reference sources
• Support integration of LLM and data workflows with FDA-furnished Elsa/HALO capabilities and approved data services, including PostgreSQL and/or HALO/Databricks, for evidence caching, analytics, retrieval, and dashboard-ready outputs
• Develop SQL and Python data access logic and assist with APIs, JSON/XML processing, data validation, and automation supporting extraction, historical retrieval, comparison, and source traceability
• Execute data- and AI-focused testing, including checks for completeness, mapping consistency, retrieval quality, unsupported or hallucinated content, ambiguity, reproducibility, prompt behavior, and regression across approved configurations
• Document data models, prompts, mappings, test results, interfaces, transformations, implementation decisions, and technical limitations for client review, knowledge transfer, and iterative POC refinement
• Work with senior data engineers, clinical protocol SMEs, data standards specialists, architects, and AI/prompt engineers to translate approved requirements into working POC components
• Participate in prompt and LLM testing, technical working sessions, POC demonstrations, validation activities, and iterative refinement based on client reviewer feedback
• Support implementation within provided platforms and approved data environments; this is a proof of concept effort and does not require independent ownership of a production platform
• Escalate ambiguous clinical, semantic, or AI-generated results to designated SMEs instead of making independent scientific or regulatory interpretations
• Build practical data and AI engineering skills while increasing expertise in clinical protocols, USDM/M11 standards, semantic models, and regulated AI validation

Requirements

• 3-5 years of experience in data engineering, software development, data science, AI/ML, analytics, or a related technical role
• Hands-on experience with Python for data processing, automation, API integration, or AI/ML workflows
• Data modeling experience and working knowledge of relational databases, XML, and JSON structures
• Practical exposure to generative AI and large language models, including prompt development, LLM APIs, structured outputs, evaluation, or application development using modern AI frameworks
• Must be able to obtain and maintain a High-Risk Public Trust (or equivalent federal client access)
• Must have the current ability to obtain and maintain required access without interruption aligned with long-term program needs
• Must be authorized to work in the United States without employer sponsorship now or in the future

Technologies

• Python, JavaScript Object Notation (JSON), XML, SQL, ETL/ELT, PostgreSQL
• Elsa/HALO, HALO, Databricks
• ICH M11, Clinical Data Interchange Standards Consortium (CDISC), Unified Study Definition Model (USDM)
• RAG, embeddings, semantic search, vector indexing
• Large language model (LLM), APIs, REST APIs, Git/version control

Preferred Qualifications

• Working knowledge of SQL and relational databases such as PostgreSQL, Oracle, or SQL Server; exposure to data modeling, ETL/ELT, metadata, or data-quality practices
• Experience working with JSON/XML, REST APIs, Git/version control, and cloud or data platforms is beneficial; exposure to AWS, Databricks, or similar platforms
• Familiarity with RAG, embeddings, vector search/vector databases, semantic retrieval, or grounding techniques; experience can come from professional, academic, or personal projects
• Relevant certifications: Databricks Fundamentals, Generative AI Fundamentals, or AI Agent Fundamentals

Location

• Hybrid in Columbia, MD

Clearance

• Must be able to obtain a High-Risk Public Trust

Education

• Bachelor’s degree in computer science, data science, computer engineering, or related field