From a classroom in Gujarat to building 19 live AI services. The unedited story — 2013 to now.
Where obsession with numbers turned into an obsession with systems.
Numbers weren't just grades — they were proof that systems, when studied carefully, reveal their patterns. Physics and Mathematics weren't subjects. They were the first programming languages.
Something clicked in a classroom. Not just the equations on the board, but the idea that everything — from how planets move to how markets behave — follows a logic. That if you understand the rules deeply enough, you can build anything.
“Physics taught me the universe has rules. Code taught me you can write new ones.”
College. Computers. The realization that building things is more powerful than studying them.
Started engineering. The classroom was familiar — equations, logic, problem sets. But something outside the classroom was more interesting: a laptop, an internet connection, and the question: what can I actually build?
Somewhere between lectures and late nights, Python stopped being a coursework language and became a thinking tool. Built scripts that automated financial models — investment analytics, banking calculations — the kind of number work that used to take hours in Excel. As Vice President of the Tech Society at MNIT Jaipur (2017-18), turned that energy outward: organizing hackathons, mentoring juniors, proving that code wasn't just for assignments. It was the year the question changed from 'what should I study?' to 'what can I automate?'
Someone paid for a script. Not a lot, but that wasn't the point. The first freelance projects — automation work, data processing, small tools that saved clients hours — proved something important: the things you build at 2 AM have real-world value. It was validation that changed the trajectory. Code wasn't a skill on a resume anymore. It was a product.
“The first time someone pays for what you built, coding stops being a skill and becomes a superpower.”
Stopped treating code as a skill. Started treating it as a superpower.
B.Tech from NIT Jaipur in hand, January 2019. By June, standing on the floor of JSW Steel Ltd as a Junior Manager — not writing code, but leading 25+ people in the operation and maintenance of 144 Coke Oven Batteries. The math degree met the real world: optimizing furnace efficiency, managing shift rotations, hitting yield targets. Delivered a 1.9% efficiency improvement that looked small on paper but meant millions at scale. It was systems thinking applied to a factory floor — and the first proof that understanding patterns works everywhere, not just in a terminal.
Left JSW Steel in August 2020 and came home to Rajkot. No job, no plan B — just a conviction that code could trade markets better than intuition. From September 2020 to February 2021, built an open-source Python library for algorithmic trading from scratch. Designed 20+ strategies with secure broker-server connections, pioneered when no tutorials or guides existed — every solution figured out alone. Built a trend probability model using parameter funnels and weightage systems, with self-improving strategies that optimized in real time. In the middle of it all, scored CAT 98.8 percentile — proved he could do an MBA, and chose not to. Then Sarthee (Feb-Aug 2021): Tableau developer building sales dashboards for senior stakeholders, delivering cost reductions of roughly 3.5 million. The grind era — isolation, obsession, and the quiet confidence that comes from figuring out hard things alone.
Joined Spinny in September 2021 as Senior Business Analyst, and walked straight into the '1-to-10 and 10-to-100 journey' — the kind of growth that breaks every manual process you have. Defined key parameters across 25+ variables and 20+ KPIs, then built interactive dashboards that became the operating layer for supply and sales analytics. Deployed data-driven strategies that drove 5x growth in the top 8 markets within a year. Managed performance for 200+ agents, wrote 400+ SQL queries, built 25+ dashboards, and automated everything possible with Python, SQL, and App Script. Delivered 61.5% cost reduction (28L saved), 3x car procurement, and pushed efficiency from 150 to 190 leads per month — a 26.7% increase. This was the chapter where automation skills met real scale, and the answer to every problem started with 'can we automate this?'
From projects to products. From code to systems. From ideas to infrastructure.
An AI-native debt resolution platform — and the first time everything converged. Not just analytics, not just dashboards, but owning the entire product end to end: frontend, backend, both teams. Full Product Lead. This was the pivot point where data skills, operations thinking, and code all merged into a single role. No more analyzing someone else's system. This was building one from scratch.
Joined Agrim as Senior Business Analyst and spent twelve months proving that data-product thinking works at scale. Led warehouse stock and price automation — pushed dispatch success from 89% to 96%, saved 20+ ops hours weekly, implemented FIFO-based pricing with 100% price accuracy. Built a dynamic EDD rule engine with buffer predictions from historic data and A/B testing that improved NPS. Created a 17L+ retailer lead database from scratch for post-Series B marketing campaigns. Financial reconciliations saving 10L+ monthly, cutting 25+ manual hours per week. Left March 2025. The last time someone else's name was on the company.
Rented a VPS. Installed Docker. Deployed a service. Then another. Then another. What started as a hosting decision became the foundation for a private AI fortress — 24+ containers, 19 subdomains, all self-owned.
Named it. Defined it. Data. Operations. Tech. Strategy. Not just a brand — a framework for how to think about building AI systems that actually run in production. 19 live subdomains. 24+ Docker services. Autonomous.
When your life itself becomes a system. Private AI as the moat. Every byte, owned.
Left Agrim in March 2025. No next job lined up. No safety net. Went full-time on CredResolve's voice AI projects — applying AI to debt resolution conversations — and ZeroOne. This was the moment the story changed. Not a freelancer picking up gigs. Not a consultant billing hours. An independent builder, betting everything on the conviction that the systems he'd spent years understanding could now be the systems he owned. Every role before this was practice. This was the real thing.
Built a multi-channel AI orchestration layer — WhatsApp, Telegram, Webchat — all routing through Gemini models, self-hosted, zero third-party data. The AI works for me. Not for anyone else.
View Project →An agentic coding platform with real PTY terminals, sub-agents, and AI-driven development. The IDE reimagined — not for everyone, built for the way I think.
Published 8 Claude Code skills and 28 GSD commands — tools for the AI developer community. Building tools that build other tools.
View Project →
This page. This system. Everything — projects, trading, knowledge, infrastructure, identity — in one place. The Life OS is not a product. It's what happens when you stop renting your life and start owning it.
Every system is a living thing. It evolves. New chapters are written in production, not in planning docs. What comes next is already running somewhere on a VPS in the cloud, waiting to be named.
“Build systems, not projects. Projects end — systems evolve.”
“The next chapter is already being written.”
← Back to the OSStudent. Dreamer. Numbers obsessed.
Overall