I'm Emmanuel King Christopher, a yooung software developer from Nigeria and the creator of NEKOVA — an AI-native programming language.

I started building it shortly after I started learning to code, in October 2025 and started to push my work to Github in june 2026 — not to learn, but because I watched people around me drop out of programming when the tools got hard, and because every language I saw treated AI as something you import, not something the language actually speaks. NEKOVA is my answer to both problems: a language simple enough to teach with, and native enough to AI that there's no SDK standing between an idea and working code.

NEKOVA is built on one belief: AI should be a keyword, not a library.

# This is valid NEKOVA. No imports. No boilerplate. No setup.
let name = ask("What's your name? ")
remember "user" as name

let result = think "Write a short poem for {name}" as text
speak result

every 1 day:
    let mood = ask("How are you feeling today? ")
    think "Give {name} encouragement based on: {mood}" as text

Every row on the right is a language keyword. Every row on the left is a dependency you have to find, install, configure, and keep updated. That gap is the entire thesis of NEKOVA.

• AI-native from the ground up — think, speak, listen, imagine, shape, watch, every aren't stdlib functions, they're grammar.
• Multi-provider AI, zero configuration — Anthropic, OpenAI, and Gemini are auto-detected; switch providers without touching your code.
• Reliability built in, not bolted on — retry/fallback blocks, observe telemetry, and mock think for testing are first-class language constructs, not patterns you have to invent yourself.
• A real database, a real web server, in the language itself — SQLite and HTTP routing (route, serve) with no framework to learn first.
• Security as a language feature — sandbox strict: gives isolated, resource-limited execution of untrusted code, natively.
• Self-hosting, in progress — NEKOVA's own lexer is written in NEKOVA, and verified token-for-token identical against the reference implementation, including tokenizing its own source. Very few languages — let alone a language built by one person — ever reach the point of writing themselves.

pip install nekova-lang
nekova run myapp.nk

Licensed under BUSL-1.1 — free for personal use, learning, and commercial products under $1M/year in revenue; converts to Apache 2.0 four years after each release.

The comparison above is where NEKOVA already beats the alternative. The direction is to make that gap wider, not just deeper:

• Agent-building as the easiest entry point in any language — tool definitions with zero boilerplate, multi-agent composition as syntax, and safety limits (capability scoping, budgets, turn limits) as defaults instead of something a developer has to remember to add.
• Hallucination mitigation as a measurable, honest feature — grounded citation-checking, verify-and-retry, and cross-model consensus, exposed as real signals a developer can act on, not a false promise that hallucination goes away.
• Full self-hosting — the parser and interpreter, written in NEKOVA itself, completing what the lexer already started.
• An education layer — because that's where NEKOVA came from, and it's staying part of where it's going.

Full technical roadmap, versioned and dated: ROADMAP.md →

NEKOVA isn't a project I built once and left. It ships continuously, and every release note is public in the repo — real fixes, dated, with the reasoning behind them, not a marketing changelog.

Languages I work in: Python · JavaScript · TypeScript · Java · SQL · NEKOVA (creator)

Technologies: React · Node.js · REST APIs · SQLite · Git · GitHub

I'm interested in:

• Programming language design and compiler engineering
• AI-native developer tooling
• Open source collaboration
• Backend systems and architecture

If you're building something interesting in AI, compilers, or developer tools — reach out.