Ryland Wittman

React Native Mobile App

Leading a Native Code Developer Team in the Age of AI Coding

Abstract: 90% of the way through developing our frontend code, I became increasingly frustrated at the heaps of mismatched code dialects accumulating in the repository. Too many pull requests contained code with the personality of the AI coding assistant the developer used. Despite implementing clear policies and standardized guidelines early on, the bigger the codebase grew, the more diverse the LLM-assisted PRs became. Instead of addressing this issue by limiting the use of AI to a single provider, or by outlawing the use of AI all together, I took an approach that turned out to be massively beneficial for the team. I forked the repository and refactored the entire frontend to use a Feature-Sliced Design (FSD) architecture. Not only did this produce positive outcomes in the universal Interpretability of AI assisted coding, it also created a very intuitive architecture flow, thereby enabling easier onboarding. We also found that it became smoother for multiple developers working on adjacent code simultaneously, and the hierarchal architecture that inherently demands strict adherence became somewhat self-regulating.

Constraint-Invariant Spatial Inference System (CISIS)

Independent Applied Analysis

Why I Built It: The Frosty program addresses the challenge of sensing in environments where the received electromagnetic field is strongly modified by turbulent, dispersive ionospheric propagation. In these conditions, assumptions fundamental to traditional radar and passive sensing—stable waveform structure, recoverable coherence, known polarization, or reliable reference signals—are frequently violated. This motivated an approach that treats dispersion and decorrelation not as errors to be inverted, but as unknown operators under which only certain physically grounded properties can remain invariant.

Abstract: CISIS is a Frosty-compatible signal processing framework for detection and tracking under dispersive, anisotropic, and reference-limited Arctic propagation conditions. The goal is to identify and exploit properties of the physical environment that remain invariant in received electromagnetic field measurements, even as waveform structure, polarization, and propagation decorrelate. CISIS operates directly on baseband I/Q measurements and does not rely on waveform coherence, reference–target correlation, or explicit channel inversion. By enforcing physically grounded spatial, temporal, and multi-site constraints, CISIS enables robust sensing in regimes where conventional radar and passive noise radar approaches fail, while remaining compatible with controlled noise-like illumination and endogenous ambient fields envisioned under Frosty. The constraints include: Spatial Consistency, Temporal Continuity, Cross-Frequency Invariance, Multi-Site Agreement, Background Subspace Exclusion, and Reference Independence. Initial implementations use Python and NumPy with clear isolation of computationally intensive kernels (covariance accumulation, subspace estimation, spatial scoring, and path inference) to support straightforward transition to C++/Rust or GPU acceleration.

LiminalGenix

Why I Built It: I started rethinking how we approach problems, tracing back thousands of years to design a mathematical algorithm based on deterministic chaos-theory with three states. Bioinformatics inspired me along the way, and I found uses in ancient DNA and rare diseases.

Abstract: Developed a web-based GUI using Python 3.11 Streamlit, modeling DNA loci with a chaos algorithm (Anchor, Liminal, Drift states) inspired by Thomson's Lamp paradox. Built MLOps on GCP with Terraform IaC, GKE, Cloud Build CI/CD, PyTorch for ML, Biopython/cyvcf2 for VCF handling, and Gosling.js for visualization. Processes ClinVar VCFs to output FASTA sequences; tests show 48.70% variant overlap (12.30% pathogenic) on 1000 variants, reducing errors by 35% and compute by 20%. Containerized with CUDA 12.1 GPU support, with a roadmap for broader genomic applications.

Water Infrastructure Optimization

Why I Built It: Working in agriculture business operations, I saw water delivery systems struggling to meet demand. I wanted a design that could double the daily amount deliverable for vital operations.

Abstract: Built a water reserve quick-fill station system while working in agriculture business operations, analyzing existing infrastructure to identify inefficiencies. Focused on structural improvements to double water delivery capacity, with a design that lowers maintenance requirements.