mkulyma — quietly wiring sovereign infrastructure

I’m an engineer and systems architect building at the intersection of AI, blockchain, wireless networks, and climate tech. My work is aimed at the Global South: energy, connectivity, and economic rails that people on the ground can actually own.

I like turning into obvious realities.

StoicSystems architectScientist at heartMeditator
Limuru, Kenya
AI · Blockchain · 5G / O-RAN · Climate tech
Scroll to begin

About mkulyma

I'm mkulyma, an engineer and systems architect who likes to live where complexity gets interesting: AI agents wrapping testbeds, blockchains quietly powering real economies, RAN stacks bringing connectivity to places that never had it, and climate infrastructure that actually works on the ground. I spend my time designing private 5G and O-RAN labs, building Cosmos-based chains, and teaching AI to be a partner instead of a toy.

My compass doesn't point to more tech for its own sake. It points to the Global South. I care about rural electrification, community energy, water infrastructure, and local compute. I want a kid in a village in Kenya to have access to the same quality of digital and economic infrastructure as someone in a global capital — and for that infrastructure to be programmable, transparent, and, as much as possible, owned by the communities it serves.

Outside the diagrams and terminals, I'm a stoic and a meditator, dying every day as an ambassador of Christ. I think in terms of kaizen — continuous, ruthless refinement. I play guitar, I'm slowly building a sound-proof room, and I'm growing a life and a marriage I'm proud of. My life is basically a lab: test ideas, refine principles, document, repeat.

Life Story

Silent observer

Foundations

I grew up as the silent kid who liked to watch how things worked. Circuits, code, people, markets — it all felt like systems to be understood. Math and engineering weren’t just school subjects; they were tools for decoding reality. I learned early that real power comes from understanding fundamentals and staying calm when everyone else is noisy.

💡 Systems everywhere At some point I realised people, markets, and radios all obey patterns. Different domains, same instinct: find the system, then bend it.
Finding the stack

Deep systems

Over time I went down the rabbit holes: networks, 4G/5G fundamentals, RF basics, and how spectrum actually behaves. At the same time, I discovered blockchains — not for speculation, but as programmable economic infra. Cosmos SDK, EVM, precompiles, real-world assets. I began treating complex systems as laboratories for decision-making under uncertainty.

Why it matters

Climate & sovereignty

The more I learned, the clearer the gap became: the Global South has talent, but lacks infrastructure. So I started focusing on climate and sovereignty. Decentralized energy rails grew from that: tokenized solar projects with real MRV, and data rails for water and utilities. The goal: let communities own a bigger share of their futures.

AI as a collaborator

Agentic era

When large language models and tools like MCP appeared, I didn’t see chatbots. I saw collaborators. I began wrapping my stacks — NIST O-RAN testbeds, chains, dev workflows, knowledge graphs — behind MCP servers and agent orchestration. AI moved from a thing I query to something that helps me run labs, distill standards, and act as a lab supervisor. My work shifted into AI + RAN + chains as one integrated system.

Billion lives

Horizon

Looking forward, I see a world where private 5G, AI, and blockchains quietly power villages, factories, schools, and labs across the Global South. Where local developers can plug into sovereign compute, sovereign AI, and sovereign energy without asking for permission. My target is simple and crazy: to touch a billion lives through the systems I help design and deploy. One lab, one network, one protocol at a time.

Core Domains

AI, agents, and infrastructure

AI as infrastructure and collaborator, not just output.

MCPAgentsKnowledge graphsRAGTooling

Blockchain and programmable economies

Cosmos-based chains that coordinate real-world value with invisible UX.

Cosmos SDKCosmos EVMPrecompilesOraclesRWA

Telecom: private 5G, O-RAN, AI-RAN

AI-enhanced private 5G and O-RAN testbeds for real-world deployments.

Private 5GO-RANAI-RANsrsRANNIST TN 2311

Climate tech and Global South infrastructure

Energy, water, and utility rails for communities usually left out.

Climate techEnergyWaterMRVGlobal South

Markets and decision systems

Markets as systems for studying uncertainty and decision-making.

Decision systemsUncertaintySignalsRiskFeedback loops

Flagship Projects

01

Decentralized energy rails and MRV

Tokenized solar and transparent MRV for community energy in the Global South.

What it is

A platform for decentralised energy in the Global South. It combines tokenized solar projects, verifiable MRV, and community-focused financial rails so that rural and low-income communities can deploy and scale energy infrastructure with transparent investor participation.

Why it matters

Rural electrification is usually slow, opaque, and underfunded. By combining GIS, IoT, tokenization, and AI-driven analytics, this platform gives both communities and investors a clear view of what’s being built, how it’s performing, and how value is shared. It’s a template for how climate tech, blockchains, and data can unlock capital where it’s needed most.

Cosmos SDKCosmos EVMNFTs / RWAGISMRV pipeline
02

Private 5G / O-RAN lab blueprint

NIST TN 2311-inspired automation for bringing up private 5G on a single workstation.

What it is

A structured approach to bringing up a private 5G lab using open-source stacks and a single workstation, aligned with ideas from NIST TN 2311. It includes scenario recipes, lab supervisors, and MCP-wrapped tools to orchestrate testbed actions.

Why it matters

Most private 5G and O-RAN work happens inside big vendor labs. This blueprint makes serious experimentation possible for teams in the Global South and small orgs, letting them go from "one box in the lab" to campus and mission-critical deployments.

srsRANOAIOpen5GS / Free5GCKubernetesMCPNIST TN 2311
03

Cosmos EVM + precompiles for real-world assets and compute

Stateful precompiles as bridges between Solidity contracts and native modules.

What it is

A Cosmos-chain design where precompiles act as stateful bridges between Solidity contracts and native modules like staking, governance, oracles, and job registries. It lets EVM smart contracts orchestrate real-world assets and compute workloads.

Why it matters

Most chains stop at token transfers. This work turns the chain into programmable economic infrastructure that can track solar fields, compute jobs, or credit risk — all while hiding blockchain complexity behind simple app flows.

Cosmos SDKCosmos EVMGo precompilesSolidityOracles
04

Knowledge graph and RAG with 3D visualisation

Sovereign knowledge engines for governments and enterprises.

What it is

A sovereign knowledge graph platform that ingests documents — from telecom standards to policy papers — and turns them into interactive, AI-queriable graph views. It combines RAG for retrieval with D3 visualisation and a Svelte-based frontend.

Why it matters

Most organizations in the Global South either underuse their knowledge or outsource it to closed platforms. This tooling lets them own their own intelligence layer, with graphs that can be explored visually and queried via AI.

RAGLLMsD3.jsSvelteVector search
05

Decision lab and DRL experimentation

Complex signals as a live lab for decision-making and agents.

What it is

An evolving platform for building, testing, and improving decision-making agents over noisy, real-world signals. It includes agentic analytics, bots interfaced through chat platforms, and rigorous tracking of performance and risk.

Why it matters

Decision-making under uncertainty is one of the hardest problems in engineering and life. This lab is a playground for AI decision-making and a discipline engine for designing, deploying, and refining robust strategies.

PythonDRL frameworksExchange APIsBacktestingAgent frameworks

Lab & Methods

My work is defined not just by what I build, but how I build it. I operate with a "lab-first" mentality—treating every project as an experiment in a controlled environment.

In the age of AI, I view LLMs as active collaborators. My workstation is an ecosystem of MCP servers, knowledge graphs, and agentic workflows that amplify my output and ensure rigor.

Lab-first

process

Every serious idea starts in a lab. One workstation, one repo, one testbed. If I can’t bring it up in a controlled environment with clear steps, it’s not ready. From there I scale: from lab to campus to mission-critical deployments, always with observability and rollback paths.

Agent-augmented

process

I treat LLMs and agents as collaborators. I wrap my stacks behind MCP servers — chains, RAN testbeds, knowledge graphs, data pipelines — and let AI handle the repetitive tasks: scaffolding, search, summarisation, experiment orchestration. My job is to design the system and decide what matters.

Principle-led

values

Stoicism and faith give me constraints. Not every shiny problem is my problem. I choose work that aligns with sovereignty, climate resilience, and long-term value for the Global South. That means saying no to a lot of noise so I can go very deep where it counts.

Documentation as leverage

process

I like turning chaos into reusable structure. PRDs, whitepapers, diagrams, meta-prompts, lab recipes — these are tools so that future me and other builders don’t have to start from zero. A good document should function like an API: clear, composable, and easy to build on.

Initiate Uplink

If you're building AI, networks, climate tech, or sovereign infra for the Global South—or you just want to talk labs and systems—reach out.

I prefer conversations that lead to experiments, deployments, or long-term collaborations.

© 2025 mkulyma

Limuru, Kenya