JAHANZAIB ALI

I build the infrastructure behind AI agents — from the agent runtime and LLM gateway down to the hosting layer it all runs on.

View Projects Get in touch

SCROLL TO ENTER

$ whoami

About

I build production-grade systems solo — end to end. That means the AI agent runtime and its orchestration layer, the LLM gateway it talks to, the database and infra underneath, and the hosting stack that keeps it all online. Not prototypes — systems with admin control planes, observability, budgets, and audit trails from day one.

My work spans autonomous computer-use agents, self-hosted LLM routing and proxy infrastructure, fintech platforms handling real transactions, and the hosting/licensing systems underneath. Python and Rust on the backend, Next.js and NestJS on the front, Postgres and Docker everywhere in between.

Full-stack, solo-built AI agent infrastructure Self-hosted LLM systems Production, not demos

$ ls ~/skills

Skills

Languages

Python TypeScript Rust PHP

Frameworks

Next.js NestJS FastAPI LangGraph / LangChain

Data & Infra

PostgreSQL Redis Prisma Docker Turborepo

AI & Agents

MCP RAG Multi-agent orchestration LLM routing

$ gh api /users/Jahanzaib211

Activity

··· Public repos

··· Followers

··· Member since

See full activity on GitHub →

FLAGSHIP · LIVE

NovaThe Agent's Computer

Problem

Most AI coding agents are a black-box chat window — you can't see what they're doing, or trust the result, until it's already done.

Approach

Nova gives the agent an actual computer: a live, streaming six-facet view — Files, Terminal, Editor, Browser, Activity, Review — so you watch it work in real time, with sub-agents, memory, and per-thread sandboxes handling execution.

PythonLangGraphLangChainNext.js

nova.alilabsx.com → Code

01 / 04

Forge Studio

Problem

Running production AI features across multiple LLM providers means juggling separate API keys, no shared budget visibility, and no failover when a provider goes down.

Approach

A self-hosted proxy layer in front of every provider — smart routing with automatic fallback, per-provider budget tracking, a visual proxy-pool topology with live health/latency, and a Prometheus /metrics endpoint so every request is observable.

TypeScriptNext.jsPrometheus

alilabsx.com → Code

02 / 04

Genesis X

Problem

Buying access to bonds, real estate, or private equity as an individual investor usually means opaque intermediaries with no unified view of pricing or holdings.

Approach

Genesis X packages these instruments as digital products investors browse, buy, and manage directly — transparent pricing on top of a full transactional backend: auth, payments, and holdings.

Next.js 14NestJS 10PostgreSQLPrisma

alilabsx.com → Code

03 / 04

Ali Kernel

Problem

Swapping the LLM behind an agent usually means rebuilding the tooling, memory, and safety scaffolding around it from scratch for the new backend.

Approach

A single-binary Rust gateway in front of any OpenAI-compatible backend — tool access, RAG memory, semantic routing, and a self-critique loop stay constant no matter which model sits behind it, from a local llama.cpp server to any hosted provider.

RustRAGSemantic routing

Code

04 / 04

fox-hosting

Problem

WHMCS is the industry-standard hosting/billing platform, but it's locked behind an external license server and an ionCube-encoded API that's slow to integrate with a modern frontend.

Approach

Pairs WHMCS 8.10 with a custom-built license server (no external whmcs.com dependency) and a Next.js 16 client portal that talks to MySQL directly, bypassing the encoded API entirely.

PHPNext.jsDocker

Code

$ open contact

Let's build something.

GitHub LinkedIn alilabsx.com