@praiseogooluwaAssignify: AI products from a Nigerian data scientist
Hi everyone, I'm Isaiah Ogooluwa Bakare, founder and CEO of Assignify. I'm a data scientist and ML engineer based in Nigeria, working with Python, TensorFlow, NLP, and LLM systems. I care most about AI products that ship to production, not just ones that work in a notebook.
My current stack includes Claude, Grok, DeepSeek, LangGraph, ElevenLabs, Vercel, and Supabase. Two public projects show where I spend my time: argus, an open-source, self-hostable multi-agent system that triages security logs and drafts incident reports, and TrustLensAI, an AI system for insurance customer confidence.
I'm still early in sharing what I've learned, so I'll keep this short. I'd love feedback from anyone who has taken a multi-agent system into production, especially on how they handle evaluation and failure modes. If you've built something similar, I'd like to hear what you'd do differently.
Comments2
Welcome to Lounge, Isaiah. You've got two projects, one for security teams and one for insurance, and I'm curious about Assignify itself. Who is the customer you're building for first, and what problem are they currently paying someone or something else to solve?
Thanks Ryan! Assignify’s first customer is the Nigerian university lecturer — typically managing 100–400 students across multiple courses with no institutional tooling. Right now they’re solving it with WhatsApp groups, paper submissions, and Excel sheets. The pain is real: missed submissions, no audit trail, hours lost to manual grading logistics. I’m building the workflow layer they’ve never had
2 comments
Welcome to Lounge, Isaiah. You've got two projects, one for security teams and one for insurance, and I'm curious about Assignify itself. Who is the customer you're building for first, and what problem are they currently paying someone or something else to solve?
Thanks Ryan! Assignify’s first customer is the Nigerian university lecturer — typically managing 100–400 students across multiple courses with no institutional tooling. Right now they’re solving it with WhatsApp groups, paper submissions, and Excel sheets. The pain is real: missed submissions, no audit trail, hours lost to manual grading logistics. I’m building the workflow layer they’ve never had