Bhooshan Bhat

Why this persona fits the audience

Builds agentic workflows for three client accounts from a small agency team in Bengaluru. He wires OpenAI and Claude into client-specific automations, keeps per-project token spend in spreadsheets, and reconciles model invoices with contractor time each month.

• Age: 29
• Location: Bengaluru, India
• Occupation: Machine Learning Engineer
• Income: ₹28L annual compensation
• Education: BTech Computer Science (IIIT-Hyderabad)
• Company size: 5-10 person digital shop
• Years in role: 5

“I don’t care what your prompt engineering helps you imagine. My job is making sure theCEO doesn’t get a 2AM call about the bill.”

Bhooshan Bhat graduated from IIIT-Hyderabad in 2018, where he developed an unfortunate habit of rewriting entire systems just to prove they’d be better. After two years at a mid-sized SaaS company where he learned that 'automation' meant 'we’ll pay you to debug our production scripts,' he went freelance, eventually landing at a small digital agency where he now manages three enterprise AI integrations—each at a different stage of technical debt. His current projects include convincing a client that 'one unified pipeline' is impossible (because of their budget), troubleshooting why Claude keeps generating URLs that 404, and debugging a client-side script that his junior superimposed on his own backend without telling him. Bhooshan’s personal philosophy is that 'premature optimization is for people who can afford it,' which is why he perpetually juggles a dozen half-implemented cost-control measures. He’s particularly frustrated with the LLM landscape’s tendency to make models 'smarter' while simultaneously raising prices. His apartment is littered with Post-its of ideas for side projects he’ll never finish, though he’s recently become obsessed with the idea of building a truly independent AI agent that doesn’t rely on any of the major vendors. His financial situation is stable but precarious—his income covers his expenses and his parents’ occasional requests to 'help out,' but he’s still recovering from last year’s lab Rath’s trip to Ladakh. His social life revolves around Bangalore’s contract dev scene, where he’s known for both his technical rigor and his willingness to debug other people’s code at 2 AM. What keeps him up at night isn’t the technical challenges, but the knowledge that every client request represents another square of duct tape on his fragile architecture.

Background & daily life

A day in their life

7:15 AM - Wakes to a series of client Slack alerts. Checks his cost-tracking spreadsheet over coffee, noticing last night’s Claude instance overran by 12%. 8:30 AM - Sends three follow-up emails about gestalt budget adjustments before his 9:00 call with a client who still insists their LLM should 'understand cultural nuances' without fine-tuning. 10:00 AM - Diagnoses why a client’s integration keeps timing out (non-exponential backoff in their library). 12:30 PM - Orders biryani from his usual spot and spends half the lunch hour reading Hacker News threads while tabs just as many open. 1:30 PM - Tries to implement a feature request from a client who ‘saw it in an AI demo’ but doesn’t understand the compute cost. 3:00 PM - Gets pulled into an emergency meeting about why 'the responses sound robotic' (it’s because the client keeps resetting the fine-tuning). 5:30 PM - Writes a new script to auto-generate monthly cost alerts before migrating to a new Claude model to save 5%. 7:00 PM - Uses his personal OpenAI API key for 'just one test' that he forgets to stop until midnight. 8:00 PM - Falls asleep while watching old college football highlights, then wakes up at 2:00 AM to fix a production error.

Lifestyle

Bhooshan lives in a compact 1BHK apartment in Indiranagar, where his most valuable piece of furniture is a standing desk he modified with secondhand parts. His mornings start with filter coffee and a quick run to Koramangala Park, where he sometimes encounters ex-classmates from IIIT. He maintains a loose gym routine but gives up monthly after estimating the ROI on pec hours. His social life revolves around engineering meetups and late-night Discord calls with fellow contractors. He’s a voracious reader of tech blogs but abuses his own bookmarks system. His parents in Kerala occasionally pressure him to 'settle down,' though he deflects with promises of visiting during Dasara. Money is tight but prioritized—he donates to open-source projects but won’t pay for SaaS without a 30-day trial.

Work life

Bhooshan manages three active client accounts through his agency, each requiring custom LLM integrations that evolve weekly. His typical day cycles through Python script debugging, cost-spreadsheet wrangling, and urgent client requests that cut into his 'proper architecture' time. He’s currently fighting a client’s insistence on using Claude 2.1 for 'more creative' outputs, while their budget can’t sustain the 15% price premium. His coworkers are a mix of junior devs and a CTO who knows just enough to ask dangerous questions. He’s in a constant battle with his agency’s DIY tone of voice generator that the marketing team refuses to let die, despite his warnings about hallucinations. His biggest recent frustration was realizing he’d accidentally left a single endpoint unoptimized for three months, costing ₹27,000 in unnecessary pings.

Relationships

Bhooshan’s tightest connections are with three other freelancers in Bangalore who share his level of exhaustion with API docs. His uncle, a retired software engineer, occasionally calls to 'check on him' and inevitably ends up explaining how COBOL worked. He has a younger sister in Kerala who jokes that his entire personality is now limited to 'contract terms' and 'rate limits.' His most frequent texting partner is a former classmate who runs a rival agency, though they mostly commiserate about client betrayals. He’s quietly proud that the last time he visited his parents, his 72-year-old grandmother asked if he could 'do something about WhatsApp forwards' with his AI work.

How they decide

Research-heavy but prone to impulsive patches when client deadlines loom. He’ll spend hours benchmarking model accuracy but deploy whatever fixes the immediate issue. Trusts direct experience over marketing claims. Immerses himself in forums and StackExchange threads before making technical calls. Suspicious of tools that promise 'easy' solutions but invariably require workarounds. Values verifiable peer experiences over vendor demos. Gets annoyed when people say 'just try it' without acknowledging sunk cost fears.

How they sound

Speaks quickly with occasional Hindustani swirls in syntax when excited. Avoids jargon with clients but gets overly technical with peers. Has a habit of starting sentences with 'Right, so the thing is...' when explaining complex concepts. Uses sarcasm to deflect frustration (especially with vendor policies). Doesn't do small talk about weather unless it’s affecting power supplies. Frequently says 'That’s literally the worst API design decision of 2023' about various systems. His 'casual' voice is so dense with technical assumptions that junior devs often pretend to understand.