Kelvin Roman
4K posts

Kelvin Roman
@KelvRoman
Software Engineer · Building cool stuff

Users from South Africa, Kenya, Tanzania, Nigeria, and the Mauritian diaspora in Europe have found Babel organically. No ads. No marketing budget. No influencer campaigns. They found us and stayed. One user sent over 100 messages in a single day! 1/2

Introducing MedScan - our first medical imaging model. MedScan classifies brain MRI scans into glioma, meningioma, pituitary or no tumour with confidence scoring. Built from scratch by @DarajaAI, it’s the first step in building comprehensive AI-assisted diagnostics for Africa.




Babel now has real-time internet access across all 15 languages. Ask about Formula One results in English. Check oil prices in Kreole Morisien. Get the latest world news in Swahili. Research any topic in any language Babel speaks. daraja.ai/chat #AskBabel #AIforAfrica





La gente que sabe *de verdad* de inteligencia artificial lleva diciendo esto desde bastante antes de que los LLMs fuesen populares. De hecho, la gente que sabe *de verdad* de inteligencia artificial afirma que aún faltan entre 50 y 100 años para que cualquier sistema artificial llegue al nivel de inteligencia de un perro. Otra cosa es que simule que es inteligente o que sus capacidades sean sobresalientes, pero eso no es inteligencia.

Kickstart your coding journey Join our Introduction to Programming with Python course and gain practical, job-ready skills. 🗓 11–27 May 2026 ⏰ 5:30–8:00 PM (Weekdays) 📍 Virtual | 💰 Ksh 5,000 Open to students & professionals. 🔗 smartacademy.go.ke/introduction-t…

as someone who works on making LLMs run faster and cheaper every day, i can confidently say the question of whether theyre conscious has exactly zero impact on whether theyre useful. we dont need our inference stack to be conscious, we need it to be correct, fast, and affordable. the consciousness debate is fascinating philosophy but its a distraction from the actual engineering problems that determine whether AI creates value. the gravity formula doesnt need to exert weight to help you build a bridge @Hesamation

Google DeepMind researcher argues that LLMs can never be conscious, not in 10 years or 100 years. "Expecting an algorithmic description to instantiate the quality it maps is like expecting the mathematical formula of gravity to physically exert weight."


Babel-Giza average response time: 8.5s. The average Babel translation confidence: 83%. Search-augmented query accuracy: real-time, sourced, cited. Languages live: 15. Countries covered: 32+. Days since launch: 15. We ship daily. daraja.ai/chat #AIforAfrica






