Self-taught. I learn by building things from scratch and sharing the work in public.
I'm a self-taught developer from India. I build backends, mobile apps, and AI-powered features, and I'm also learning how AI works under the hood instead of only using ready-made libraries.
I believe in understanding the tools I use, so I try to build the core ideas myself first and then compare them with the standard libraries.
🌐 Portfolio: hariomlohardev.vercel.app
I'm currently learning AI and machine learning step by step, in public.
- Done so far: math foundations (linear algebra, calculus, probability) and neural networks from scratch with NumPy (backpropagation, optimizers)
- Learning now: deep learning — CNNs, attention, Transformers, and PyTorch
- Up next: building a small language model, RAG at scale, and more research-style projects
My code, derivations, and notes live in AGI_Research.
- AGI_Research — my learning notebook: ML and deep learning concepts built from scratch
- spam_classifier — a Naive Bayes spam classifier trained on 5,572 messages
More projects on my portfolio → hariomlohardev.vercel.app
Building with
Strong foundations
Currently learning
Harvard CS50P — CS50's Introduction to Programming with Python (2026) 9 problem sets + final project · Verify certificate →
I'm available part-time for:
- Backend Systems — Django and FastAPI APIs, authentication, databases
- Mobile Apps — Flutter and Dart, one codebase for iOS and Android
- AI Features — LangChain and RAG over your own data
- Automation & Scripts — scraping, pipelines, and Python tools
📩 Reach me through my portfolio or email hariomlohar.new@gmail.com
"I don't just use the tools — I build them, so I actually understand how they work."
© Hariom Lohar · hariomlohardev.vercel.app