John Thomas johnthotekat

Hey, I'm John Thomas

Technical Architect @ Prevalent AI | Creator of Data Bridge

I wrote my first line of C++ in 2003 and never really left the screen since. I'm not the guy who grinds LeetCode for fun - I'm more the guy who needs to know how things work under the hood, and occasionally ends up building them.

I was one of the early ones who jumped into Big Data back in 2013-14 - built production-grade Cloudera Hadoop clusters, managed them through Hue, and watched the ecosystem grow from MapReduce jobs to Spark to everything running on Kubernetes today. Those were good times, and I still miss them. But what excites me is how far we've come - from wrestling with YARN configs to orchestrating GPU workloads on K8s.

Over the years, I've worked across multiple organizations building engineering teams and products from the ground up. Currently, I'm the creator, architect, and product owner of Data Bridge at Prevalent AI - an ingestion platform that can extract data from virtually any source, including custom APIs, into the data lake.

Lately, I've been deep in the Claude Code rabbit hole - and it's been a ride. Some things I've been experimenting with:

Building RAG pipelines and document intelligence systems
Spinning up AI agents with multi-agent orchestration
Writing custom MCP servers and tools
Running local LLMs on my rig with Ollama
Grammar-constrained generation - making small LLMs follow strict output formats using GBNF grammars
AI image generation with ComfyUI, IPAdapter, and consistent character workflows
On-chain analytics and automated trading strategies on Solana
Full-stack apps - from data pipelines to frontends, all vibe-coded with Claude

The Rig

My daily driver for all of this:

CPU    AMD Ryzen (B650M Gigabyte GAMING X AX)
GPU    NVIDIA RTX 3080 Ti (12GB VRAM)
RAM    32 GB DDR5
SSD    NVMe
OS     Ubuntu Linux (kernel 6.17)

Good enough to run local models, ComfyUI workflows, Docker stacks, and the occasional CS2 session.

The Journey

Era	Years	What I Was Building
The Foundation	2003 - 2009	C++, Java, SOAP/XML Web Services, servlets, JSP. Learning how things work from the ground up.
The Enterprise Years	2010 - 2013	Spring, REST APIs, Java EE, Oracle, production-grade backend systems.
The Big Data Wave	2013 - 2017	Cloudera Hadoop clusters, Hive, MapReduce, Apache Spark, Kafka. One of the early ones in - built it before it was cool.
The Platform Era	2017 - 2021	Python Django, Spring Boot, microservices, Docker, data pipelines at scale. Building engineering teams and products.
The Cloud Native Era	2021 - 2024	Kubernetes orchestration, cloud-native data platforms, CI/CD, FastAPI, everything containerized.
The AI Era	2024 - now	RAG systems, AI agents, MCP servers, local LLMs, on-chain analytics, full-stack apps - all vibe-coded with Claude. The stack matters less when you understand the architecture.