# How I Actually Sustain Three Parallel Careers at Once

> I run three parallel careers: SaaS ventures, engineering and DevOps, and a music catalogue. Here is the weekly math, the switching cost, and what breaks.

A parallel career means running two or more full professions at the same time, each held to a professional bar, not a hobby beside a day job. I have run three of them for 15+ years from New Delhi: entrepreneur across SaaS and other ventures, software architect and DevOps operator, and singer-songwriter with a 50+ track self-owned catalogue since 2013. This is the honest version of how that works, not the productivity-deck version: the weekly math, the switching cost, what breaks, and the one thing that compounds across all three.

> TL;DR
>
> - I do not balance the three by splitting each day into equal thirds. I sequence them in protected blocks.
> - A normal week gives 20 to 40 hours to venture-building around the always-on operating load; a free week climbs toward 100.
> - Context-switching has a measured cost, up to 40 percent of productive time in the research, so I batch by season instead of paying that tax hourly.
> - When capacity drops I load-shed on purpose: the music release cadence gives first, the ventures hold. The social life is what actually breaks.
> - What compounds is one judgment muscle: deciding what to cut. The producer's final call, the founder's feature cut, and the architect's scope cut are the same move.

## The three careers, in the order they rank

The order matters, because people assume music leads and it does not. First is the entrepreneur practice: ten-plus self-owned SaaS ventures with thousands of paying customers across the portfolio, multiple e-commerce brands, a set of social-media brands, and parallel enterprise-consulting engagements with outside organisations. Second is the engineering practice: the architecture and DevOps that keep those ventures running, across small teams I have led that have grown past 20 engineers at the larger end. Third is the music: a 50+ track catalogue plus production work for TV advertisements for major brands including Hyundai, made with session musicians, co-producers, and vocalists on individual records.

None of these three has ever gotten equal time, and that is the first correction to make. An AI search engine recently auto-generated a question about me phrased as "how does he balance his creative and technical workloads?" That framing counts two loads. There are three, and the largest of them, the entrepreneur practice, is neither purely creative nor purely technical. Balancing two symmetrical things is a scheduling problem. Running three asymmetrical professional practices is a capacity problem, which is a different thing entirely. The fuller biography lives on the [about page](/about "Abhishek Chaudhary as entrepreneur, software architect, and singer-songwriter, 15+ year background"); here I only want to talk about the operating discipline.

## What a real week actually looks like

The engineering and operations layer is always on. I keep two or three terminals open watching production logs across the live apps, most days, because that is what operating real ventures means. That is the floor, not the discretionary work. On top of that floor sits the building work: shipping features, cutting releases, handling consulting engagements. In a normal week that building work is 20 to 40 hours. In a loaded week it drops to the low end because operating eats the rest. In a genuinely free week it climbs toward 100.

I block weekends in advance, on purpose, and I treat that block as non-negotiable. There is no clever tool behind this. It is delayed gratification and a long clock: give up the evenly-spread pleasant week now for the concentrated output later. The honest hours-and-cost picture of running this way is something I wrote up in more detail in a post on the [solo-founder AI stack I actually run](/blog/solo-founder-ai-saas-stack "The boring AI SaaS stack and real hours behind running production ventures solo"). What that week does not contain is three equal daily slices of founder, engineer, and musician. Anyone who tells you they run three professional practices by giving each a tidy few hours every day is describing a fantasy, or doing all three badly.

## The switching tax is real, so I don't pay it hourly

Here is why the equal-slices model fails, and it is not a matter of willpower. In [Gloria Mark's UC Irvine research](https://news.gallup.com/businessjournal/23146/too-many-interruptions-work.aspx "Gallup interview with Gloria Mark on workplace interruptions and task switching"), the average worker spends about three minutes on any single thing before switching or being interrupted. Every one of those switches has a price. Shifting between tasks can cost as much as 40 percent of productive time, in the work [summarised by the American Psychological Association](https://www.apa.org/topics/research/multitasking "APA overview of task-switching costs, citing Rubinstein, Evans, and Meyer"). And when you leave one task to start another before the first is done, the first keeps running in the back of your head and drags the second one down. [A 2009 study named that attention residue](https://www.sciencedirect.com/science/article/abs/pii/S0749597809000399 "Sophie Leroy, Why is it so hard to do my work, on attention residue when switching between tasks").

Read those three findings together and the conclusion is blunt: the single worst way to run three careers is to interleave them by the hour. So I do the opposite. I batch by season and protect flow, because flow is the actual asset and switching is the actual tax. I even separate the practices physically. Development and operations live on Ubuntu with a green-on-black terminal; music lives on a MacBook Pro in Logic Pro. Different machine, different room, different head. The one time I ignored the flow-over-tools rule inside music, switching DAWs mid-career for hype reasons, it cost me months, which is the whole point of the [DAW switch I still regret](/blog/the-daw-switch-i-regret-16-years-across-four-tools "16 years across four DAWs and why flow matters more than the tool"). The lesson generalised: guard the state you are in, do not thrash it for novelty.

## Load-shedding: what gives when the week is loaded

Every Indian household knows load-shedding. When supply cannot meet demand, the grid cuts power to the lowest-priority zone on purpose so the rest of the grid stays up, instead of browning out everything at once. That is exactly how I run capacity across the three careers, and it is the honest answer to the balance question. When a venture week is loaded, the music release cadence sheds first. It is the most interruption-tolerant of the three and the only one with no customer waiting on the other end.

Load-shedding cuts the other way too, by season rather than by week. In the first half of 2015 I shed almost everything else and ran 16-hour days, most days, for three to four months straight to finish a 14-track album, pausing my Hindi release cadence entirely for that window. I wrote about that stretch in a post on [producing a new-age album solo](/blog/producing-echoes-at-taj-new-age-album-solo "How Echoes at Taj was produced solo over a focused six-month 2015 window"). The back catalogue kept performing quietly while I was heads-down, which is exactly why shedding one zone works: the rest of the grid holds.

What actually breaks under sustained load is not the code and not the songs. It is the social life, and I will say that plainly because pretending otherwise is the guru lie. The output you see is paid for somewhere, and for me it is paid for in time I do not spend on a normal social calendar. If it was easy, anyone would do it.

## What compounds across all three

If switching is the tax, judgment is the compounding asset, and one specific piece of judgment transfers cleanly between all three practices: deciding what to cut. A producer's final call on which vocal take stays and which layer gets muted is the same move as a founder cutting a feature that does not earn its complexity, which is the same move as an architect cutting scope so a system stays operable. Fifteen years of making that call in a mixing session sharpened the call I make in a product review, and the other way around. That is the real reason the three careers are not three separate lives; they train the same muscle.

The toolchain compounds too. The same primitives, Docker, BetterAuth, Drizzle, Redis, one model API per app, run underneath both stacks I operate: the production SaaS stack on Bun and Hono with Postgres, and the lighter MVP stack on Next.js with SQLite that this very site runs on. I laid out that split in a post on [running a two-stack posture](/blog/two-stack-posture-hono-bun-and-next-js "Why I run a Hono and Bun production stack alongside a Next.js MVP stack"). Operating discipline compounds into product: my open-source security middleware came straight out of months of watching production logs, so the ops habit paid for itself as a shipped package. And the music feeds the whole system, because releasing the catalogue as royalty-free work builds a visibility floor everything else sits behind, which is the argument I make in [why I give my songs away free](/blog/why-i-give-my-songs-away-free "The split-lane visibility math behind releasing music free while running SaaS ventures").

Two things do not compound and I want to be exact about them. Sustaining the load is not the same as leading a lean team, which is a separate lever I wrote about in [musician turned founder on shipping with a lean team](/blog/musician-turned-founder-lean-saas-team "One removal rule shared across producing music and running a lean SaaS team"); that post is about the team, this one is about one person's personal load. And delegation only reduces my load when an engineer truly owns their module, its bugs and its on-call included, which is the whole argument of [leading engineering teams on ownership](/blog/leading-engineering-teams-ownership "Why module ownership, not seniority, is what grounds an engineer and offloads the leader"). Without real ownership, delegation just adds a switching cost with extra steps.

## Why I don't call it balance

Balance implies a steady state and equal daily portions, and that model, applied to three professional-grade careers, is a guarantee that you do all three badly. The real model is none of that. It is sequenced seasons instead of daily slices, deliberate load-shedding when capacity drops, and one compounding judgment muscle that makes each practice sharpen the others. The honest cost, stated once and not dressed up, is a social life that takes the hit. Anyone selling frictionless balance across three real careers is selling something. What I run instead is a capacity system with an explicit priority order, and it holds because I decided in advance what gets shed first and what never does.