smndiaye
HomeServicesWorkWritingResumeDakkantalContact

Serigne Mbacké Ndiaye

AI systems engineer · AWS · Tokyo

Taking on select freelance projects

I design and build AI systems on AWS for companies that need them to work in production: retrieval pipelines, AI agents, and the serverless platforms they run on. Ten years of engineering experience, from hands-on delivery to leading programs of thirty engineers. Based in Tokyo, working remotely with teams in English, French, and Japanese.

Start a projectSee my work

What I do

AI systems on AWS

Retrieval-augmented generation, AI agents, and LLM integrations built on Bedrock and Lambda. Designed for production: observable, cost-controlled, and owned by you.

Full-stack product delivery

From idea to deployed product: Go and TypeScript, React PWAs, offline-first apps, payments, auth. The kind of systems in my work directory, built end to end.

Cloud and serverless architecture

AWS infrastructure as Terraform code: serverless by default, near-zero cost at rest, no ops team required. Reviews and rescues of existing stacks included.

Engineering leadership

Fractional tech lead or engineering manager for your team: architecture reviews, technical hiring, and coordination of distributed teams. Ten years of delivery behind every decision.

Selected work

Qlearn

Active
qlearn.smndiaye.com

Quran learning application with reading, recitation audio, translations, memorization tools, and progress tracking.

React PWA
Go on Lambda
DynamoDB
offline-first

ai@smndiaye

Self-hosted
Self-hosted

Self-hosted terminal coding agent on open-weight models via a local vLLM server.

Go
Terminal UI
vLLM
MCP

Suñu Shop

Active
shop.smndiaye.com

Multi-tenant e-commerce platform. A mall landing lists independent shops, each with its own storefront under the same domain, plus an admin app for owners to manage products and orders. OTP and Google sign-in, with usage metered daily for pay-as-you-go billing.

React
Go on Lambda
DynamoDB
Cognito

Archives

Active
archives.smndiaye.com

Web application for browsing and listening to a podcast collection from 2003–2013. Features advanced search, custom playlists, and a modern HTML5 audio player.

React
PWA
S3
CloudFront

How I work

  1. Intro call

    We scope the problem and whether I am the right fit.

  2. Proposal

    Fixed scope, timeline, and price in writing.

  3. Delivery

    Working software in weekly increments, your repository, your infrastructure.

Remote-first from Tokyo. English, French, Japanese. NDAs welcome.

Have a project in mind?

Start a project