sudo-rndm.ai is a veteran-owned (VOSB) AI/ML research, development, and consulting firm delivering trusted, high-impact solutions for government agencies, defense partners, research institutions, healthcare partners, and small-to-mid-sized businesses. We specialize in data science, intelligent systems engineering, and advanced analytics that help organizations innovate, operate efficiently, and make confident, data-informed decisions.
Research we're building
Flagship R&D efforts where we apply our methods to hard, high-stakes problems.
polar navigation / ● in dev
fig. — a route that respects the ice
Boreas Navigator
Calibrated, multi-modal deep learning for sea-ice characterization, short-horizon forecasting, and uncertainty-aware polar route planning.
- •Multi-modal retrieval across SAR, passive microwave, and optical imagery
- •Conformal prediction sets for distribution-free uncertainty
- •Uncertainty-aware, risk-bounded route optimization
lightning nowcasting / ● in dev
fig. — strike probability on a grid
Bronte Nowcaster
Calibrated, sub-hour deep learning that predicts lightning strike probability at each point on an airfield or launch range — refreshed every 5 minutes, out to 30. A live demo is running now, replaying real archived storms.
- •Attention-ConvLSTM-Conv U-Net over satellite, radar, and lightning feeds
- •Calibration-first: probabilities you can base a go/no-go call on
- •Software-only, satellite-based — deployable to data-sparse sites
military justice / ● in dev
fig. — evidence mapped to elements
Themis
Privilege-aware evidence synthesis and case-preparation for military justice — bounded, auditable, and human-in-the-loop, with a formal confidence guarantee on every output.
- •Evidence mapped to the elements of a charge, traceable to its source
- •Conformal selective prediction: abstains and routes to human review
- •Privilege-by-design, on-prem, no frontier-model dependence
$ sudo rndm --list-capabilities
Our capabilities
Six practice areas, one standard: calibrated, explainable systems that hold up in high-stakes environments.
fig. 01 — model, prototype → fielded
AI/ML Research & Engineering
explainable ai · edge ml · llm pipelines
End-to-end model development, from prototype to fielded system — built for real-world operational conditions.
02fig. 02 — forecast, calibrated bounds
Data Science & Decision Analytics
bayesian methods · forecasting · optimization
Bayesian inference, forecasting, and statistical modeling for mission planning and operational decisions.
03fig. 03 — many modalities, one estimate
Multimodal Systems Integration
sensor fusion · remote sensing · computer vision
Image, video, sensor, text, and structured-data pipelines with cross-domain fusion for remote sensing and analytics.
04fig. 04 — updates move, data doesn't
Federated & Privacy-Preserving Learning
federated learning · privacy by design · edge deployment
Secure, decentralized learning frameworks for multi-site data environments where the data can't move.
05fig. 05 — white paper → funded program
Proposal and R&D Support
sbir / sttr · white papers · prototyping
SBIR/STTR teaming, white papers, technical writing, and rapid prototyping for innovation programs.
06fig. 06 — one team taught, many fielded
Education, Training & Outreach
workshops · curricula · upskilling
Tailored AI/ML training and upskilling for government, industry, and small-business teams.
Our team
Chance Johnstone, Ph.D.
Co-Founder & CEO
Ph.D. in Statistics (Iowa State). Twelve years as a U.S. Air Force data scientist and former Assistant Professor of Statistics at AFIT. Expert in uncertainty quantification, optimization, and decision science for high-stakes environments.
Get in touch
Tell us the problem you are trying to solve. Your message goes straight to a founder, and you will get a real reply.
- chance@sudo-rndm.ai
- Location
- Centerville, OH