Michael Asomugha (Lyon)

Social media can really make or break you 😂 I stumbled upon an IG ad by @saintkpogode for the CEO & Founder Gathering – Lagos at Radisson Blu VI, shared it with my execs, and we showed up.

It’s another edition. #hangoutwithmike

Banks and ads when my account Dey on red 😭

Run a full coding agent locally. No API bills. No limits. No data leaving your machine. Private. Powerful. 100% free. I made a step-by-step guide anyone can follow in minutes. To get it, just: → Like + Repost → Comment “LOCAL” → Follow me (so I can DM)

@folaoftech AI is for urgency 😂😂

Software developers 5 minutes before the deadline 😭

if I perish, I perish. 42k 😭😂. #relivethemagic     #makeithappen     #Lagos42km    #runLagos     #accessbanklagoscitymarathon         #RunYourRaceLagos         #run42k

🤷🏽‍♂️🤷🏽‍♂️ linkedin.com/posts/michael-…

@TheObiLeonard I’ve been thinking of getting one. Sounds like you have an experience. I’d love to hear.

Every tech bro will go through that phase where they believe that purchasing an iPad will improve their life, work, and career significantly, please don’t let that person be you. Book a vacation trip. Get a pet. Start a book club. Anything else but that iPad.

3/3 Fix checklist: 1. Beat strong baseline first 2. Plan deployment + monitoring Day 1 3. Align on business metric (not just AUC) 4. Build cross-team ownership (DS + Eng + Biz) ML wins = engineering + ops + problem fit > fancy architecture.

2/3 Top killers when model is good: - Data issues → poor quality, drift (fashion reco from 2020 fails in 2025) - No MLOps → no monitoring, no retraining → performance dies silently - Wrong problem → solves something nobody needs or can’t measure ROI on

1/3 Even a “perfect” ML model (99% accuracy in notebook) often fails in production. Real stat: ~70-85% of ML projects never deliver real value (McKinsey/Gartner reports). Why? Model is only ~10-20% of success.

Awesome Stuff.

#ML #PythonTipsPick

Pro move instead of fixed val set? → Use cross-validation on train_full

How pros split data in scikit-learn 2026 style: First , hold out final test set (never touch it until the very end)

The Goal: Learn the best strategy (policy) to maximize rewards over time. Example: AlphaGo learning to beat the world champion at the game of Go.

Reinforcement Learning 🎮 Learning by trial and error. An "agent" observes an environment and performs actions, receiving "rewards" 🍬 for good moves and "penalties" ⚡ for bad ones. The Vibe: Training a dog with treats.

Machine Learning generally comes in three distinct flavors. 🍦🤖 Here is the breakdown of Supervised, Unsupervised, and Reinforcement Learning. 🧵👇