AI Engineer

About Us

At Verisian, we build deep tech and AI solutions that enable groundbreaking medical therapies to enter the market faster and safely. New medicines and devices are judged using all available evidence in the proper context and backed by automated analyses that validate and make data and results transparent to companies and regulators alike. Regulators will make the right decisions on novel therapies faster and with greater confidence, protecting patients from harm and making breakthrough treatments available as soon as is possible and safe.

In times of increasing trial numbers and complexity, we are building technical innovation to remove crucial bottlenecks for pharmaceutical and medical device companies, as well as public health authorities, directly affecting clinical trial planning, analyses, validation, submission, and market approval.

By joining us, you will commit yourself to building software and AI tools that directly contribute to increasing the rate at which medical innovation improves human health and wellbeing.

Disclaimer:

We welcome candidates of diverse experience levels, and meeting every requirement is not mandatory. Research indicates that underrepresented groups tend to apply only if they meet all qualifications. If you're enthusiastic about the role, please apply and let our recruiters evaluate your application.

Culture

At Verisian, our mission is to build the future infrastructure of medical innovation. To help us succeed, we're creating a unique employee culture. We always put the mission first. We're fanatically customer obsessed, crafting world-class products that customers love with every interaction. We take extreme ownership and accountability of our work, seeing whatever we do through to completion. We communicate candidly and directly with each other, even when it's uncomfortable. We're innately curious, open to alternative perspectives and invest passionately in our own continuous growth.

Role Description

As an AI Engineer, you will join our world-class engineering team in building the Verisian Platform. You will use the latest in AI, specifically LLMs and RAG systems, in combination with Verisian’s game-changing code and data traceability to create the world’s best

The Verisian Platform brings value to a set of highly regulated processes crucial for medical progress and innovation. You will be working on the Planner, Builder, Explorer, Validator, Submitter, and related supporting modules. These core modules of the platform target the planning, exploration/onboarding, building, validation, submission, and review of clinical trials and their results. They enable data managers, statistical programmers, statisticians, medical writers, and regulators to deliver their work faster, at higher quality, lower cost, and in greater confidence.

Our pipelines analyze clinical trial documentation, code, logs, data, and results to build a knowledge graph through code traceability. We harness the resulting dataset, column-level and logic lineage to turn clinical trials into Information Infrastructure that can be used by experts and consumed by AI to revolutionize how therapies are evaluated and enter the market. We capture complex processes in fully- and semi-automated workflows that place experts in control and AI automation at their fingertips. We build visualizations to provide our customers with a maximum of insight as fast as possible.

Our application stack is based on Next.js and deployed via Docker/Kubernetes in the cloud. The data analysis pipelines run in Argo Workflows. We analyze code based on Antlr4 and Java. AI agents are developed in Python. The data analysis engine is developed in Python. Git is where our code lives, and Github Actions is how it gets out into the world.

You will be expected to lead the analysis, design, and engineering of our AI stack (RAG system, selection and fine-tuning of models, production deployment), develop strategies for development, regression testing, training/validation data and associated benchmarks, and contribute to its production-ready deployment. As part of our core team, you will join us in designing, prioritising, building and testing new functionality, troubleshooting customer issues, finding root causes, and developing improvements to ensure maximal user impact and performance.

Coding Agent Role:

This role will focus on the Verisian Agentic AI Coding Agent. You will use specifications, code, data, and documentation from existing trials accessible, pre-structured, and consumable through our APIs to build agents that plan and then execute statistical analyses on either existing or new trials.

Verisian’s code traceability allows for the retrieval of exactly and only those lines of code required to derive specific variables/outputs as well as exactly and only those lines of code that are affected by any introduced changes. This means our agents can retrieve the perfect context for planning the correct and side-effect free changes and deciding exactly where to make them, no matter the size of the code base. This solves one of the fundamental problems of modern coding agents: accuracy as complexity and size increase.

For new analyses, they will extract code templates from previous related studies and refactor/modify them in accordance to updated requirements. Changes to existing studies are direct modifications on the trial’s code base. The Coding Agent changes will run autonomously in SAS, R, and/or Python. Its outputs are then validated using the Verisian Validator, our semi-automated validation system to ensure accuracy, consistency, and confidence before a submission. Apart from generating the analyses, the Coding Agent will resolve issues identified in our Validator autonomously.

Analysis/Results Validation Role:

This role will focus on the Verisian Agentic AI Validation. You will use specifications, code, data, and documentation from existing trials accessible, pre-structured, and consumable through our APIs to build agents that test if the data and statistical results are calculated according to their specifications.

Verisian’s code traceability allows for the retrieval of exactly and only those lines of code required to derive specific variables/outputs, providing the perfect context for comparing the implementations in code to their natural language specifications, no matter the size of the code base.

The agents will report differences between specification and implementation and make suggestions for how to fix them to ensure an integral analysis and subsequent submission.

• Experience with and foundational understanding of LLMs (especially open source models), including production deployment
• Experience with and foundational understanding of non-LLM AI, including production deployment
• Experience with RAG systems
• Strong interest in programming languages, parsing algorithms, interpreters, and compilers
• Strong interest in understanding data science and statistical analyses
• Extensive experience in Python, Pandas, and at least one other programming language
• Highly motivated and highly independent, able to create and manage a code base
• Strong focus on building as a team: we have a We & Mission mindset
• Strong ability to communicate complex technical problems and solutions, foresee risks, and align work across multiple teams and colleagues
• “If it’s not code and documented, it doesn’t exist”-mindset
• Iteration mindset

Bonus Points for:

• Experience with code generation/validation/analysis
• Experience in data/statistical analysis
• Experience in clinical trials

• What you build impacts billions of people around the world
• Highly collaborative, ambitious and world-class team
• Employee Stock Options Plan
• All remote, asynchronous work environment with in-person summits around the world
• Pension plan and additional benefits depending on country of residence