LF

Machine learning, cloud and backend engineer, and founder of Vikolo Dynamics. I design and ship complete systems, from models and APIs to cloud infrastructure and native apps.

Hire me
  • Cloud
  • AI
  • ML
  • Product

Machine learning

Speaker embeddings, live Training 00%

A simulation of how a voice is matched to its nearest neighbours.

Capabilities

Cloud
AWS, Google Cloud, Firebase, Cloudflare, Vercel
Languages
Python, TypeScript, JavaScript, Swift, Kotlin
Backend
Node.js, Next.js, FastAPI, Docker, REST and streaming APIs
AI / ML
LLM APIs, speech-to-text, embeddings, model serving

The path

  1. Systems

    Networks, service management and cloud architecture.

    • AWS
    • ITIL
    • Networking

  2. Products

    Founded Occupating, then native apps, web and APIs.

    • LLMs
    • iOS
    • Android
    • Web

  3. Machine learning

    Founded Vikolo Dynamics to build ML products.

    • Python
    • Embeddings
    • Serving

Featured · RecitID by Vikolo

  • Swift
  • Kotlin
  • TypeScript
  • Python
  • Serverless
  • Machine learning
Contact

Available for hire.

Machine learning, AI features, cloud and APIs, native apps.

LinkedInGitHub

About

Portrait of Luqman Folorunsho

I’m Luqman Folorunsho, an engineer and founder. I started in infrastructure, moved into building products, and machine learning now sits at the centre of what I make.

I founded Vikolo Dynamics, a technology company building machine-learning systems, the services that run them and the apps on top. RecitID is our first product.

I like owning the whole system: the data, the model, the cloud it runs on, the apps people hold and the details they notice.

Certifications

  • AWS Solutions Architect
  • ITIL Service Manager
  • CompTIA Network+

How machine learning became part of me

It started with one question and turned into a way of working.

  1. The question

    Could a machine tell who is speaking from the voice alone?

  2. The data

    Recordings collected and cleaned, filtered and loudness-normalised, so the model hears the voice and not the room.

  3. The representation

    A speaker-embedding network turns a few seconds of audio into a point in space. Each voice gathers in a neighbourhood of its own.

  4. The iterations

    Version after version, tested against real recordings, until the neighbourhoods separated cleanly.

  5. In production

    Deployed as its own service and called in real time.

RecitID

Every layer, built end to end: native clients, an API, a machine-learning service and the cloud underneath.

Client
Native iOS and Android apps.
API
A serverless API that prepares requests and coordinates the services behind it.
Model
A machine-learning service that recognises voices.
Data
Managed authentication, database and object storage.

Built with Swift, Kotlin, TypeScript and Python on serverless infrastructure.

recitid.ai

Capabilities

Cloud architecture and backend engineering, with machine learning built in.

Cloud

AWS
Solutions Architect certified. S3 and CloudFront static hosting.
Google Cloud
Serverless containers on Cloud Run, Cloud Build CI, Artifact Registry, scheduled jobs.
Firebase
Authentication, Firestore and Cloud Messaging push.
Cloudflare
R2 object storage at zero egress.
Vercel
Next.js sites and publications.

Backend

APIs
Node.js / Next.js and Python / FastAPI services with token auth, validation and rate limits.
Realtime
Streaming responses over server-sent events, with cancellation.
Integrations
Webhooks, push notifications, scheduled jobs.
Operations
Containers, CI/CD, telemetry, budget alerts, circuit breakers, rollbacks.

Languages and more

Languages
Python, TypeScript, JavaScript, Swift, Kotlin
AI / ML
LLM APIs, speech-to-text, speaker embeddings, model serving
Clients
SwiftUI, Kotlin / Compose, React
Service
ITIL service management, networking

The path so far

  1. Systems first

    I started underneath the software: networks, service management and cloud architecture. It taught me to build things that stay up, can be observed and recover on their own.

  2. Then, products

    I wanted to make the things people hold. I founded Occupating to help people position their careers with language models, then built native apps, the web and the APIs behind them.

  3. Then, machine learning

    A product needed a model that could recognise a voice. Building it changed how I work, and led me to found Vikolo Dynamics: I now start from the data, train and serve the model, and ship the product on top.