Bala Sivagnanam

AI is not magic. It is the same software architecture it has always been — how systems connect. What changed is that LLMs now handle the repetitive, formulaic parts. Earlier we wrote all that code by hand. Now we delegate it. The systems thinking, the judgment, the design — that still belongs to you. Most AI coverage buries this under buzzwords. I try to say the plain thing.

I lead technology at BalanceFlo.ai, an ergonomics program management platform for companies. The work is about making ergonomic wellness easier to assess, manage, and scale with AI, especially in places where manual programs are too slow or too expensive to run well.

My path moved from application development into UX, then into company building. I have spent years turning design ideas into front-end code, internal tools, corporate websites, research reports, wireframes, usability tests, and now AI-assisted systems.

AI systems without the mysticism — architecture, agents, skills, and stacks explained as what they actually are. Personal builder tools: why everyone needs their own setup now, and how to build one that fits how you actually work. Design thinking applied to AI — how a UX lens reads agentic systems differently from how engineers or executives tend to read them.

Masala Geek is part notebook, part lab, part public memory. It is where rough research can become a note, a note can become a LinkedIn post, and a repeated question can become a project page. The site shows the work, not only the polish.

BalanceFlo, Carousel Studio, and TheBalaOS: a local-first personal builder system that connects capture, research, Markdown, Git, the public website, and social publishing without turning the whole thing into a giant CMS.

I write for e27 and publish notes and build logs here. If you are organising a talk, panel, or workshop on AI in practice, agentic systems, or building with AI — reach out on LinkedIn.

LinkedIn ↗  ·  GitHub ↗  ·  RSS ↗

Astro, Markdown, and Git. Hand-built, local-first, and deliberately small.