2015 to 2018 · Paris, France

Koalike

Founder

Koalike was my first startup venture: an Instagram growth software product that reached roughly $20k in monthly recurring revenue, with a $29.99/month average plan and customers across North America, Europe, Australia, and New Zealand.

The systems behind the product

Find the right conversation. Then engage with context.

Koalike read the language and image in each post, checked whether it fit a brand's audience, and queued relevant work. That helped brands take part in conversations more thoughtfully.

Input

Instagram activity

Captions, hashtags, images, and brand interests.

Koalike artificial intelligence

Understand each post.

Read the languageCaptions and hashtags become signals.
Word2Vec
Understand the imageImages add visual context.
Cloud Vision API
Judge the fitSentiment and relevance identify a fit.
TensorFlow + Prediction API

BigQuery + App EngineFind activity, then queue work at a measured pace.

Product outcome

Relevant conversations

Help brands engage thoughtfully with their community.

Language, image, and relevance signals guided thoughtful community engagement.

The stack in plain terms

Each tool had one job: understand posts, judge relevance, and run that work at scale.

TensorFlow

model training

A foundation for training the sentiment model in-house and keeping the scoring loop close to the product.

Google Prediction API

hosted prediction

A hosted machine learning service: train a sentiment classifier on labeled brand posts and call it as a simple web endpoint.

Word2Vec

language signals

Google's open word vectors for turning captions, comments, and hashtags into meaning the model could score.

Google Cloud Vision API

image understanding

Read image content and brand context, since an Instagram post is mostly the picture, not the caption.

BigQuery

large-query analytics

Store and query the millions of brand posts crawled each day; already part of the production analytics stack.

Google App Engine

production runtime

Task queues and Datastore that ran the crawlers and paced thoughtful community engagement at scale.

peak monthly recurring revenue

$20k

average plan

$29.99/month

customer reach

global

I sold Koalike to customers in the United States, Canada, 13 European Union countries, Australia, New Zealand, and other markets.

Work breakdown

A practical map of the systems, surfaces, and decisions behind the work.

I founded Koalike as an Instagram growth automation product for brands and creators.

I grew the product to roughly $20k in monthly recurring revenue on a $29.99/month average plan, implying around 667 active subscriptions at that revenue level.

I built workflows that analyzed account content, used tags and interests to find relevant audiences, and supported deliberate, rate-limited community engagement.

I shipped the operational software surface around it: dashboards, action toggles, interests, pricing, subscriptions, account linking, support, admin, task queues, and analytics.

I participated in the early Station F ecosystem.

Community discovery and engagement

I built Koalike to find the conversations and communities that fit a brand, using account analysis, targeting, and controlled engagement.

Production software machinery

I grew the product to roughly $20k in monthly recurring revenue with a $29.99/month average plan, backed by Google App Engine services, APIs, Datastore models, task queues, Stripe billing, BigQuery analytics, dashboards, mobile account linking, and admin/support flows.

Worldwide customer base

I took Koalike beyond a local-only experiment and into a real international subscription footprint.

Marketing systems through-line

I can trace a line from Koalike to Aranx: audience discovery, automated growth workflows, measurable acquisition loops, and building the whole product myself.