Jason B. Johnson

AI Software Engineer - Biosecurity and LLM Evaluation Focus

About Me

Welcome! I am a husband to an incredibly hardworking nurse and a proud dad of two teenage boys. I am a software engineer with three decades of product and platform delivery experience at JPL/NASA, including software on multiple Mars Rovers. I currently work at The RAND Corporation as a Senior AI Software Engineer, building LLM-enabled systems and evaluation frameworks.

I am focused on AI security and biosecurity engineering roles where I can contribute as an individual contributor on backend systems, LLM workflows, and safety-oriented technical evaluations.

I grew up on a cattle ranch in Northern California and went from cowboy to coder.

On the Ranch

Space Flight Software

My software has played a pivotal role in powering and advancing key operations of NASA's Jet Propulsion Laboratory (JPL) missions, including the iconic Mars Rovers: Spirit, Opportunity, Curiosity, and Perseverance; the trailblazing Ingenuity helicopter on Mars; as well as instrumental scientific endeavors like the Europa Clipper, the Psyche Asteroid Mission, and the NISAR climate observation satellite. By providing essential functionality across a diverse range of applications—from exploration and data collection on distant worlds to monitoring Earth's climate—this software has become a cornerstone of contemporary space exploration and research, contributing to some of the most significant cosmic discoveries of our age.

Rover Media

My involvement with the rover missions is documented in the book Roving Mars: Spirit, Opportunity, and the Exploration of the Red Planet by Steve Squyres.

You can also learn more about the rover missions by watching Good Night Oppy on Amazon Prime.

AI Research

LLM Workflow Engineering

Autograder Optimization Built a repeatable workflow for improving LLM-based essay grading by combining programmatic prompt optimization with enterprise endpoint integration.

Multi-Model Integration Integrated multiple LLM providers for production-like evaluation and benchmarking flows, with support for synchronous and asynchronous usage patterns.

Agentic Systems Designed practical agent workflows for analysis-heavy tasks with clear evaluation criteria and reproducible outputs.

Biosecurity Agent Evaluation
Designed and implemented evaluation tasks to test how well LLM agents handle gene-design challenges under realistic biological constraints. Built scoring and validation workflows to measure capability while emphasizing misuse-risk awareness. This includes White House-directed evaluation support on Llama 405B biosecurity risk.
Serious Games + LLM Platforms
Built multi-user serious game systems with real-time collaboration, adjudication workflows, and AI advisor integration. Delivered game engines and web platforms where humans and LLM agents can participate through shared interfaces.

Robotics Research

Rocky 7
Rocky 7 was an early rover prototype. My job was to create software to test rover performance. This also involved commanding the rover to drive around the JPL Mars Yard.
Robotic Assisted Microsurgery (RAMS)
Designed and fabricated custom hardware to test the fidelity of RAMS. I wrote custom software and drivers to get feedback from the robot as it interacted with the hardware. Listed as co-author on the research findings.
CHIRP Sonar
Built hardware and software to further research on CHIRP Sonar, now commercialized by Garmin.

Applied LLM Engineering

Production-oriented AI Integration
Hands-on delivery across Azure OpenAI, AWS Bedrock, and Ollama, including provider integration, scoring/autograder patterns, and human-in-the-loop workflows for complex analysis tasks.

Engineering Leadership at Scale

I have led large, complex software delivery portfolios in mission-critical environments, with an emphasis on platform reliability, product delivery, and cross-functional coordination.

Organizational Leadership: Part of the senior management team guiding an IT organization of over 700 employees and contractors, with a budget exceeding $100 million, providing strategic direction and governance.

High-Impact Communication & Collaboration: Effectively navigated complex organizational dynamics, communicating across executive, middle management, policy, and technical teams to ensure alignment and delivery.

Systems Strategy: Helped shape long-term technology strategy and architecture direction for large institutional systems with high operational expectations.

Vendor & Contract Management: Managed vendor relations and contract negotiations, ensuring the high availability of critical infrastructure. Reviewed key procurements to ensure alignment with IT roadmaps and overall strategy.

Selected Delivery Highlights

Governance Process Modernization
Modernized review and decision workflows to improve speed, alignment, and technical accountability across major initiatives.
AI-enabled Review Automation
Developed AI-driven digital twins for project and portfolio reviews, reducing manual analysis overhead while improving consistency.
Platform Rationalization
Streamlined overlapping platforms to improve maintainability and reduce duplicate operational costs across enterprise systems.

Digital Transformation Strategist

In my capacity as Digital Transformation Strategist at JPL, I led the effort to author a Lab-wide IT Strategy. Our team identified and oversaw the implementation of critical digital transformation initiatives across the entire organization. We curated a portfolio of projects, using our business case analysis to ensure alignment with JPL's strategic goals and to secure funding.

My role was pivotal in architecting the solutions, leveraging my technical acumen to guarantee their integration within our complex enterprise architecture. This task required an in-depth understanding of JPL's intricate system landscapes and technological roadmaps, as well as a keen insight into a broad spectrum of digital technologies. By strategically deploying these technologies across various business functions, we aimed to revolutionize operational efficiency, enhance data management practices, and secure mission success.

Sample Projects

Semantic Search
We dramatically reduced search times for experts, in some cases going from weeks to minutes, by leveraging graph technologies.
Manufacturing/PLM
I re-architected our manufacturing ecosystem from the ground up to seamlessly integrate our key mission data using SOA, API-First, and Event driven architectures. This dramatically reduced silos of data across the Lab.
Advanced Data Analytics
I helped evolve JPL past the paradigm of dashboards to insights, using advanced trend analysis software such as ThoughtSpot to provide real-time analytics to complex business processes.

Interim CDO and CKO

Over the span of two years, before JPL hired on a full-time Chief Data Officer and Chief Knowledge Officer, I was asked to step in to fill some of the needs in that area.

In my capacity as Interim Chief Data Officer, I worked to craft a 200-plus-page Data and Analytics Strategy for JPL. As a part of this effort, I created a maturity model for JPL systems to assess the data quality, integration readiness, and analytics potential of all of JPL's institutional data. It took a team over a year to complete the assessment. The assessment was used to prioritize which areas of JPL needed the most attention and we secured funding to advance the maturity of those areas. This assessment is now maintained under the new Chief Data Officer.

As part of the Interim Chief Knowledge Officer role, I worked with a consultant to help craft what Data Behaviors and what Data Ethics we wanted to encourage at JPL. We worked with Human Resources to understand the results of JPL culture studies and paired those insights with behavioral psychology. We rolled out a change management campaign around data behaviors that included engaging videos and an interactive game to help people learn about how to handle data and knowledge at JPL. I also crafted a proposal to help JPL evolve the Chief Knowledge Officer role to reflect industry trends.

Key Deliverables

Data and Analytics Strategy
This comprehensive strategy gives an honest assessment of the state of JPL's data, the roles required for JPL to evolve in its data management, and the practical, measurable next steps required.
Data Behaviors and Ethics
This work sought to explore: What behaviors do we want JPLers to exhibit when creating and consuming data? What are our Data Ethics best practices? What moral obligations do we have with respect to our data?
Digital Maturity Assessment
I sought to answer the question "Is JPL Digital, Programmable, and Smart?" I expanded on this Gartner framework to assess JPL across nine dimensions of maturity.

Professional Skills

Core capabilities most relevant to AI security and biosecurity-focused individual contributor roles.

LLM Workflow Engineering
Expert
Backend Software Development (C#, .NET, Python, C/C++)
Expert
System/Platform Engineering
Expert
Secure Technical Delivery
Expert
LLM Provider Integration (Azure OpenAI, Bedrock, Ollama)
Expert
Evaluation Framework Design and Autograders
Expert
Database Design / SQL / Administration
Advanced
UX and Web Design
Advanced

Education

Research and Policy Development

Part of the RAND Graduate School Master of Technology Policy inaugural cohort and co-author of Toward Comprehensive Benchmarking of the Biological Knowledge of Frontier Large Language Models (RR-A3797-1).

View RAND Co-authored Publications

B.S. Computer Engineering and Computer Science, University of Southern California

I spent my time at USC preparing for a career in robotics and artificial intelligence. I did a hardware/software combo degree, along with a minor in cognitive psychology. The intent was to understand how humans behave, learn, and sense so that I could replicate that in robots and AI.

M.S. IT Leadership and Management, Claremont Graduate University

This program mixed elements of an MBA, Technology, and Executive Leadership. I did in-depth course work in cryptography, knowledge management, and geographic information systems.

AI/ML/Deep Learning and Ongoing Study

Completed a comprehensive series of certifications through DeepLearning.AI, covering key areas of AI, machine learning, and deep learning. I gained hands-on experience in PyTorch and TensorFlow, with a focus on NLP and generative AI systems.