Building AI for Environments Where Trust Matters.

Turns mission requirements into governed AI programs that can be scoped, delivered, and defended.

Evidence-governed AI fails or succeeds long before a model is chosen — in how requirements, accountability, and evidence are defined. Connal brings 15+ years of technology, analytics, federal IT, and program leadership to exactly that work. As Senior Project Manager supporting AT&T Federal’s GSA Enterprise Infrastructure Solutions (EIS) portfolio (2019–2026), he managed data-intensive pricing and contract workflows, analyzed SOWs and solicitation requirements, reconciled large datasets for government submissions, and bridged federal stakeholders with technical teams — contributing to $300M+ in federal task-order activity and earning recognition as an AT&T Federal Lead Performer for 2025–2026. As Program Director at Soothsayer Analytics (2016–2019), he directed predictive analytics and decision-support programs from requirements discovery through delivery.

Has already built the core patterns of evidence-governed AI — grounded retrieval, document extraction, and hallucination control — inside the World Bank Group.

Jason brings 7+ years of AI and data science in a high-stakes financial institution. As a Data Scientist at IFC (World Bank Group), he built an enterprise question-answering and summarization system that connected Azure OpenAI GPT-4 to proprietary documents through LangChain, embeddings, and vector search — so investment teams received answers grounded in their own records. He implemented LLM-agent financial spreading with Azure AI Document Intelligence to extract and break down statement line items for due diligence, built a multi-agent GPT-4 policy-generation system in which a critic agent challenged outputs to reduce hallucinations, applied structured LLM outputs to map ESG report statements to UN SDGs, and engineered Databricks/PySpark Medallion pipelines that move raw data to analysis-ready form.

Builds the secure, access-controlled, auditable infrastructure that sensitive AI systems have to run on.

An AI system is only as trustworthy as the environment around it. Kairine brings 8+ years of securing and operating mission-critical cloud deployments. As an AWS/Azure DevOps and DevSecOps Engineer at Leidos (2019–2023), she automated infrastructure with Terraform and CloudFormation — reducing provisioning time and cost by 35% — built CI/CD pipelines, eliminated single points of failure across availability zones, implemented SLOs and error budgets, and secured Kubernetes workloads with Istio and Calico network policies. In her current DevSecOps role she integrates security testing into CI/CD, enforces IAM and role-based access control, leads incident response, supports system security plan (SSP) and PTA documentation, and reduced average response time by 80% through Ansible automation.