. Fielding .

Hello, I'm Fielding Johnston. With over 15 years of experience, I specialize in building scalable AI infrastructure, healthcare platforms, and distributed systems. Currently I'm building a patient communications platform at Freed. I care deeply about AI adoption, not just shipping AI-powered products, but helping the teams around me get better at using AI every day.

This portfolio highlights my work, starting with my Professional Experience. After that, check out my academic Publications and recent Projects.

See my work on GitHub, find my research on Google Scholar, connect on LinkedIn, or grab my Resume. Want to work together, collaborate on shaping AI’s future, build something cool, or just chat? Reach out at hello@justfielding.com.

.Professional Experience.

Senior Software Engineer

Freed | 2025 – Present

Founding member of a 4-person team that built a greenfield patient communications platform; selected as the company’s top strategic bet and scaled to a larger team.

Built end-to-end voice and SMS infrastructure including patient identification with dual validation, AI-initiated call transfer, two-way SMS with TCPA-compliant consent, and PHI screening via LLM.

Shipped BigQuery analytics pipeline as source of truth for pilot metrics, integrating sentiment analysis and call evaluation criteria, and fixing duration reporting accuracy from 0% to 100%.

Designed care collaboration workflows with PostgreSQL RLS and role-based access, enabling multi-clinician visit collaboration with realtime patient updates.

Created and shipped ‘12 Business Days of AI,’ a 12-day internal engineering campaign with guest contributors teaching advanced AI patterns, shifting the team from shadow AI usage to documented shared mastery.

TypeScriptReactSupabasePostgreSQLTwilioElevenLabsOpenAI RealtimeBigQuerytRPCGraphQLGKEPlaywrightGitHub Actions

Vice President of Engineering

Nesa | 2023 – 2025

Architected a distributed inference framework for 120,000 models, handling up to 2M requests per day on heterogeneous hardware.

Designed and implemented Nesa’s layer 1 blockchain for decentralized AI compute.

Built telemetry pipeline using NATS.io and Prometheus to monitor nodes.

Led secure containerization for user-submitted models.

Contributed to equivariant encryption (EE) achieving zero-latency private inference.

Developed reinforcement learning system for dynamic model assignment.

Contributed to open-source: xFusers and nats.py stability improvements.

PythonCUDANATS.ioPrometheusDockerblockchaindistributed systemstensor/sequence parallelismPyTorchLangChainLLMsHugging FaceRLHFTypeScriptBashGitLinux

Lead Software Engineer & Director of Design and Development

Opessky | 2022 – 2023

Led Unity-based lunar colony simulation game development.

Designed and implemented 3D environments and shaders.

Built AI systems for dynamic NPCs with procedural behaviors.

Directed art team and coordinated UI/UX integration.

UnityC#3D programmingshader programming (GLSL)AI (NPC behavior)game designJavaScriptproject leadership

Lead & Senior Frontend Engineer

Vidy | 2019 – 2022

Lead Frontend Engineer (2020–2022)

Refactored to Svelte, improving performance for 14M+ monthly users.

Built components for scalable real-time video and data applications.

Developed custodial wallet system for user transactions.

Implemented NLP for ad placement to boost engagement.

Senior Software Engineer (2019–2020)

Built scalable APIs and web apps for real-time video.

Led adoption of modern JS frameworks.

Engineered fine-grained reactive vanilla JS framework.

JavaScript (ESNext)SvelteTypeScriptReactHTMLCSSRollupPostCSSGitAPI developmentNLP

Founder and Software Engineer

justFielding | 2009 – 2019

Built and scaled web apps integrating React and Three.js.

Advocated open-source frameworks to reduce tech debt.

Managed technical, design, and business operations.

JavaScript (ESNext)ReactThree.jsRxJSHTMLCSSWebpackNode.jsC++NginxRedisGitDocker

.Publications.

Meta-Learning for Speeding Up Large Model Inference in Decentralized Environments
Yipeng Du, Zihao Wang, Ahmad Farhan, Fielding Johnston, et al. (2025). COLM 2025 Conference Submission.
Encrypted Large Model Inference: The Equivariant Encryption Paradigm
J. Buban, H. Zhang, C. Angione, Fielding Johnston, et al. (2025). arXiv preprint.
Model Agnostic Hybrid Sharding for Heterogeneous Distributed Inference
Claudio Angione, Yue Zhao, Harry Yang, Fielding Johnston, et al. (2024). Submitted to MLforSys2024.
Towards Secure and Private AI: A Framework for Decentralized Inference
Hongyang Zhang, Yue Zhao, Chao Yang, Fielding Johnston, et al. (2024). NeurIPS 2024 Workshop RBFM.

.Projects.

Yarrtzee: Dice of Dishonor

A steampunk pirate-themed multiplayer Yahtzee game built as a Telegram Mini App and available as a web app. Along with scratching a personal itch, I wanted to explore the potential of multiplayer games as Telegram Mini Apps and the outreach they can provide. Also, steampunk pirates!

Fite.bet

Fite.bet reinvents online giveaways by transforming them from a simple random name(s) picker into an epic 8-bit fighting game tournament between the participants' AI-generated sprite representations. I took this opportunity to explore various diffusion model pipelines for dynamic asset generation in games and apps. It's a service that prioritizes both being awesome and offering utility.

→ Visit Fite.bet → Add Telegram Bot

Rock, Paper, Scissors

An experimental UI control system that allows users to interact with web applications using real-life hand gestures via their webcam. This proof-of-concept leverages a lightweight, browser-based neural network for real-time gesture detection to control a rock-paper-scissors game without any physical input.

→ See it in action

Zuck Hunt / Zuck's Revenge

A satirical game concept exploring gesture-based controls. "Zuck Hunt" is a light-gun style shooter (a la Duck Hunt) controlled by hand gestures, while "Zuck's Revenge" flips the script into a side-scrolling beat-'em-up. Both are explorations in creative art direction and novel user interaction.

Opessky: Lunar Simulation

A lunar colony simulation game with dynamic AI-driven NPCs. This project showcases my versatility. Acting as the lead engineer and also directing the overall design and direction of the project, I had the opportunity to touch literally every aspect of the game, from shaders to the art, style and lore of the game.

→ Watch clips from development → Lunar Field Manual LFM 3-80

Redstone University

A free, hands-on online course teaching computer architecture fundamentals through Minecraft’s Redstone system. Build registers, logic gates, and an ALU in-game, making concepts like binary and bitwise operations tangible. Created for my daughter, CS students, and anyone seeking a practical grasp of low-level computing.

→ View the Course