I'm Mihir.
I build software and the systems around it.
These days that means everything from shop-floor hardware to cloud setups to the internal tools a team runs on, mostly the unglamorous parts that just have to work.
Now
Updated June 2026
I build things and follow whatever I'm curious about, fussing over the details along the way. Most days spiralling between thinking nobody knows what they're doing and realising everything around me is someone's life's work.
Exploring
Design and ideas for my hyper-personal site, gloriously self‑indulgent. Not live yet, open to ideas.
How tech can support local Indian artists, and help people choose better as buyers.
Building
Founding engineer since 2021. Architecture, infrastructure, and the SaaS products on top, from the shop floor to the cloud.
An early-stage product built on applied AI. More soon.
Enjoying
Book. Computer-science ideas as a surprisingly good guide to everyday decisions.
Rick Rubin on finding meaning in the mundane.
Blog
Posts & Projects
Trusting the Machine: Cross-Validating LLM Output With a Second Model
Language models are great at reading messy documents and giving you clean, structured data back. They are also good at getting it wrong in ways you will never spot. This is the story of how I stopped trusting a single model and started letting two of them check each other's work.
Information Extraction using Convolutional Neural Network
Text summarization is a technique of briefing a large text document by extracting its significant information. Extractive text summarization involves direct extraction of sentences from the original document to form a summarized document. The considered task had been an intriguing one for a long time and thus many approaches had been proposed for the same. This paper proposes information extraction from a large text document using a Convolutional Neural Network(CNN).