About AppliedXL
AppliedXL detects material changes in authoritative records and turns them into verified intelligence.
We monitor regulatory filings, government records, scientific data, company disclosures and other primary sources to determine what changed, whether it matters and what the evidence supports. Our intelligence powers news, research and analytics products used by financial institutions and information providers. At the center of our platform is the AXL Event Engine, which combines domain-specific event definitions, continuous monitoring and evidence verification to create structured, traceable event records.
This is an initial six-month contractor role, with the option to extend based on fit, impact and company needs.
We’d love to learn a little about how you think. Please send us a brief cover letter telling us about a domain, sector or industry you’d be excited to map, what you’d want to track, and why. We’ll only consider applications that include one. Email at support at appliedxl dot com.
The Role
We are hiring a Computational Journalist to develop and deploy the research and AI workflows behind our intelligence products. The work centers on four questions: What changed? Is it material? What evidence supports it? What does it affect?
You will help extend our system into new domains, improve how we detect and verify change, and turn successful experiments into intelligence products used by customers.
You do not need to be a software engineer. You do need to be rigorous with incomplete records, fluent with AI tools, and comfortable turning ambiguous research questions into systems that can be tested.
What You’ll Do
- Map new domains using the AXL Event Engine, identifying authoritative sources, material events and the evidence needed to verify them.
- Build AI workflows for research, extraction, classification, monitoring and analysis.
- Turn unstructured records into structured entities, events and signals with clear source lineage.
- Develop methods for detecting changes in values, states, claims, commitments, relationships and timelines.
- Distinguish material events from routine updates and noise.
- Verify evidence, reconcile conflicting information and preserve the basis for each conclusion.
- Build evaluation sets and test outputs for accuracy, omissions and unsupported claims.
- Investigate failures across source data, retrieval, model behavior and workflow design.
- Test new models and tools against real-world tasks and adopt them when they improve performance.
- Work with research, product and engineering to turn successful experiments into reusable workflows and customer-facing products.
What We’re Looking For
- Deep practical familiarity with LLMs and current AI tools. You already use them to research, analyze or build.
- Experience designing multi-step AI workflows for research, extraction, classification, monitoring or similar tasks.
- Strong analytical judgment when information is incomplete, messy or conflicting.
- Comfort with APIs, databases and structured data, even if you are not a production engineer.
- A habit of checking model outputs against the underlying evidence.
- Ability to move quickly from an unclear problem to a working, testable approach.
- Strong writing and source discipline.
- Curiosity and a track record of teaching yourself new domains, sources and tools.
Experience may come from journalism, research, intelligence, data, product, financial information, scientific analysis or another field involving complex records.
Useful Experience
- Structured extraction, entity resolution or event-based data.
- Evaluation design and systematic testing of AI outputs.
- Healthcare and life sciences, government and regulatory records, or financial disclosures.
- Retrieval, tool use or agent workflows.
- Tracking how information changes over time.
- Turning experimental workflows into systems others can reliably run.
How You’ll Work
You’ll join a nimble team working directly with leadership, engineering and our partners. Projects move quickly and often begin with an open research question rather than a predefined specification.
The approach is practical: understand the sources, build the workflow, test it against the record, identify where it fails and improve it until the results hold up. One week you might map a new domain; the next, debug an extraction workflow or turn a prototype into something a customer relies on.
