What political campaigns taught me about India

After coming back from the US, I was trying to figure out what I wanted to do next.

I had left a funded PhD in Economics because I felt disconnected from the Indian context I wanted to study. I was working on Indian data, but I could not relate to it properly. I did not understand districts, villages, local bodies, panchayats, or how government actually works at different levels.

So when I heard about IPAC, the Indian Political Action Committee, it immediately looked interesting.

Several of our IIT Kanpur seniors and juniors were already there. I connected with a few of them and got a sense of the work: meeting bureaucrats, NGO heads, village panchayats, commerce bodies, unions, party workers and politicians was part of everyday life. For someone who wanted to understand India from the ground, this looked like one of the best opportunities around.

I applied, got interviewed, and got in. The salary was half of my US stipend, but it was enough in India to repay my loan, sustain myself, and save a little. More than that, I felt I would learn something no classroom or dataset could teach me.

Starting with election data

I started in Noida, where we spent about a month looking at election data across states: vote shares, constituencies, past performance, caste equations, turnout, margins.

It gave me a real numerical handle on elections. But very soon I realised numbers alone are never enough in India. A constituency can look one way in the data and behave very differently on the ground. A candidate can look strong on paper while local people feel something else entirely. A community can appear aligned with one party historically, until local leadership, personal relationships, or one recent event changes the picture completely.

Data can tell you where to look. Ground reality tells you what is actually happening.

Punjab: learning campaign execution from the ground

After Noida, I moved to Chandigarh for Captain Amarinder Singh’s campaign in Punjab.

Initially I was on the team working on campaign design: structure, messaging, collateral, the ground implementation plan. We’d present to the core team and to Prashant Kishor himself whenever he came to Chandigarh.

After some time, I was sent to a few districts in Ludhiana and Jalandhar: about 10 days visiting two centres and meeting people across the surrounding villages.

Then I moved into campaign management for “Halke Vich Captain” and “Coffee with Captain.” Our job was to reach a district at least two days before Captain’s visit, coordinate with local party workers, arrange promotions, put up banners, check the venue, understand local issues, and make sure the event actually worked on the ground.

One important responsibility, whenever Captain arrived, was making sure he met villagers and common people rather than being surrounded only by Congress workers. That’s simple to say and not simple to do: local party workers naturally want access, local leaders want visibility, everyone wants to stand close to the candidate. But the point of these events was for the candidate to hear from common people directly, not from the people managing him. And other messaging we wanted to ensure was - to change the image from Maharaja Amarinder Singh to approachable Captin Amarinder Singh - we were aiming multiple targets with a single arrow.

I visited close to 20 constituencies and more than 300 villages in about two months. After that I was assigned Ludhiana for a few weeks, then moved to Amritsar with three other people. The four of us were responsible for all 11 rural and urban constituencies there. I stayed around three months, working with NGOs, religious bodies, opposition members, party cadre, and local people. We even built a team of 11 leads and 220 volunteers across Amritsar on IPAC payroll for various ground operational works.

Punjab taught me how much planning sits behind what finally shows up as one public event, and that politics isn’t just speeches and rallies. It’s coordination, trust, local relationships, follow-ups, small promises, local anger, community pride, and daily execution.

It changed my complete perception around politicians. In almost every constituency I was meeting hundreds/thousands of people, however, I always found politicians to be the smartest and most hard working person there. It is much more difficult than it appears to manage narratives, conflicting demands, and opinions in limited budget; along with ensuring the vote-bank dynamics.

During the same time when we were managing Amritsar, the surgical strike happened. We were staying close to the Attari border and were advised to remain indoors.

Uttarakhand: same planning, different outcome

After Punjab, I moved to Delhi along with Prashant Kishore to help organise an all-Congress adhiveshan around Rahul Gandhi at Talkatora, planned and executed in 15 days, then moved to Dehradun for the Uttarakhand campaign.

After a few days there, two of us got responsibility for Uttarkashi district - comprising of three seats: Gangotri, Yamunotri, and Purola in Uttarkashi. It was completely different from Punjab: mountainous, beautiful, and very hard to cover. Sometimes we weren’t sure if we were there for work or on vacation. But as an election state, it was much harder than it looked.

In Punjab, our planning and execution had worked even against AAP’s momentum. In Uttarakhand, the same playbook didn’t travel automatically. We could spend a full week building local momentum, and then one two-hour visit from BJP central leadership in any part of state, Amit Shah in this case, could wipe out that week of work even in a place as remote as Uttarkashi.

Every state has its own political psychology. Punjab had one kind of energy, Uttarakhand another: terrain, leadership, local networks, party strength, voter expectations, and media flow were all different. You cannot understand India by building one framework and applying it everywhere.

Uttar Pradesh: learning caution

After Uttarakhand, I moved to Lucknow for the UP elections, staying mostly across Lucknow, Pratapgarh, and Mau. These were very different regions, each with its own local power structure.

We were told to be more careful than we’d been in Punjab or Uttarakhand, so my role narrowed to providing party workers with collateral, passing issues up the hierarchy when they wanted something communicated, and reading and recording exact ground realities.

Politics has a local grammar too. What works in one state doesn’t necessarily work in another, and the same behaviour reads as confidence in one place and unnecessary aggression in the next. You need to understand local power before you act. This was one of the first times I really understood how much informal power matters in India: there’s the formal power of administration and party structures, and then there’s local influence that’s written nowhere but that everyone understands.

Andhra Pradesh: language and local context

After Punjab, Uttarakhand, and UP, we got two months of paid leave, most of which I spent resting. The previous year had caught up with me.

Then we moved to Hyderabad for YSRCP’s campaign under Jagan Mohan Reddy in Andhra Pradesh. We started by reading the numbers: groups within the state, caste dynamics, constituencies, previous results. After about a month I moved to Rajahmundry and Kakinada for a ground check.

This is where language became a real barrier. I couldn’t interact directly with normal people and villagers the way I had elsewhere, so we leaned mostly on ex-MPs, MLAs, party workers, people from other parties, industry associations, and NGOs like the Red Cross. We still gathered plenty of information, but I could feel the difference. In Punjab, Uttarakhand, and UP I could understand people directly. In Andhra, language put me a step removed, dependent on intermediaries, and that changes the quality of what you understand. If you can’t speak to people directly, your read of what’s happening gets filtered through someone else’s interests.

What I understood about India

By the time I left political consulting, I felt I’d understood what I set out to. I’d seen how people think across states, how government bodies actually interact, and the role of media, civil society, bureaucracy, local administration, party workers, and local leaders. I’d seen how villages, panchayats, districts, and states connect; how party machinery works; how centrally run campaigns operate; how events get organised and narratives get built.

Most of all, I understood that India cannot be understood only through data, books, or policy documents. You have to go to the ground. You have to sit with people and listen to what they say and what they don’t. You have to work out who holds formal power and who holds actual influence, and accept that the same sentence can mean different things in different states: caste, religion, language, geography, economics, local pride, and personal relationships all interact in complicated ways. And even after all that, sometimes you still won’t fully understand what’s happening.

Carrying it into how I read data

Before political consulting, I’d studied economics and statistics: models, data, structured analysis. That training is genuinely useful. But the campaigns taught me that data has to be grounded in context. If the numbers say one thing and the ground says another, that’s a reason to dig further, not a contradiction to explain away: maybe the data’s old, maybe the read is wrong, maybe a community is angry but not saying so openly, maybe the person giving you feedback has his own agenda.

That’s the instinct I still use. Whenever I look at data now, I ask what’s behind it: who generated it, what’s missing, what incentive the source has, what a dashboard isn’t showing me.

Those two years were intense and often overwhelming, and also one of the most important stretches of my life: villages, small towns, district and state politics, leadership visits, local conflicts, the gap between planning and a real event on the ground. It’s also where I understood why my PhD work in the US had felt incomplete: I was trying to study India before I had actually been inside it.

That’s the two-year stretch this piece covers. The longer arc, through teaching, an unfinished PhD, and eventually AI, is at From classrooms to campaigns to code.