I grew up in Auraiya, Uttar Pradesh, India, asking why some people's choices are freer than others. Economics gave me the tools to investigate that question properly.

MSc in Development Economics at Nottingham, then research work at IDS Jaipur with India's Time Use Survey data, and now a research internship at IFAD in Rome, reviewing whether rural development interventions actually work, for whom, and how.

I'm building toward a PhD in development economics. This GitHub is the evidence trail: real problems I worked on, documented honestly, including the parts that didn't go as planned.

My research sits at the intersection of gender, labour, household economics, and measurement. How we count things, what we miss when we count badly, and what becomes visible when we count better.

My MSc dissertation looked at whether rainfall variations increase domestic violence incidents against women in India. The income stress channel made sense theoretically—less rain, less agricultural income, more tension at home, more violence.

But the data told a different story.

What I expected:  Rainfall shock → lower income → more violence  ✗
What I found:     Social sector expenditure → MORE reported violence  ✓

States that spent more on health and education had higher reported domestic violence. That is not what you'd expect if spending were simply helping.

I spent a lot of time after my dissertation reading about this: the IPV literature, measurement theory, and institutional economics. And slowly I started to think that what I found wasn't evidence of more violence. It was evidence of more reporting. When women have a clinic to go to, a school nearby, people they can tell — they report violence that was always happening but previously invisible in the data. What the data captures depends on whether institutions create the conditions for it to be captured at all.

I'm developing this as an institutional visibility mechanism — distinct from institutional trust, which implies subjective confidence, and more about the material infrastructure through which hidden outcomes surface in administrative records.

Testing and exploring such types of invisibility and mechanisms properly is what I feel I would love to do even in my PhD.

📄 Dissertation code and findings

I'm not going to pretend I'm already there. But I know where I'm going, and I'm building toward it deliberately.

2023–24   MSc Development Economics, University of Nottingham           ✓
2025      Research work at IDS Jaipur and remotely                      ✓
2026      Research Intern, Office of Development Effectiveness, IFAD    ✓  (until July 2026)
2026      Python for Data Science — Stellenbosch University             ✓  (completed April 2026)
2026      QGIS spatial data — Ecospatial Lab                            ✓  (completed May 2026)
2026      Mathematics prep — calculus, linear algebra, optimisation         (ongoing)
2026      RES Annual Conference — Speed Session on Development              (6 July, Newcastle)
2026–27   Pre-doctoral or RA position                                       (applying now)
2027      PhD applications — fully funded, development economics

Every project in this portfolio is a step on that map. Nothing here is for show.

Nothing here is a tutorial or a textbook exercise. These were real problems.

skills = {
    "Stata"        : "████████████████░░  Intermediate to Advanced",
    "R"            : "████████████░░░░░░  Intermediate",
    "Python"       : "██████░░░░░░░░░░░░  Beginner-Intermediate",
    "Excel"        : "████████████████░░  Advanced",
    "LaTeX"        : "████████░░░░░░░░░░  Intermediate",
    "QGIS"         : "██░░░░░░░░░░░░░░░░  Beginner (completed May 2026)",
    "Canva"        : "████████████████░░  Advanced  (yes, I design too)",
    "LLMs"         : "████████████░░░░░░  Regular use — Claude, NotebookLM",
    "Research tools": "Rayyan (systematic review screening), VOSviewer (bibliometric mapping)"
}

Methods: Fixed-effects panel econometrics · PRISMA systematic review · Bibliometric network analysis · Large-scale survey harmonisation · PDF processing · Data visualisation · Graphic design

Currently learning web scraping (Python) and sensitivity analysis techniques through collaborative research work — April 2026 onwards

Data I've worked with: (MOSPI) (DHS Program) (World Bank Mirodata Library)

I want to be honest about how I built the IFAD extractor because I think it matters.

When the taxonomy project came up, my consultant mentioned that Python had been tried before for this and hadn't worked well. The plan was to do it manually. I said I'd try to make it work.

I worked through it with Claude. We tried the automatic approach first — reading the table of contents, finding the page numbers. It worked for some files, but missed pages in others. I suggested trying a manual backup approach where you just specify the start and end page yourself. That worked better. We kept testing and refining.

The decisions about what to build, how to approach it, and when to try a different strategy — those were mine. Claude helped me write Python I was still learning. I also used NotebookLM and Gemini at different points to understand concepts.

I think that's honest. And I think knowing how to use AI tools well — when to use them, how to direct them, how to check what they produce — is a skill worth naming, not hiding.

My consultant's response when it was done:

"Your work has been a great help to us. I'm sure when we move to other replenishment cycles, your Python code will bring even greater value."

That was nice to hear. But what I liked most was that the tool will keep working after I leave.

→ See the code

Python for Data Science
Stellenbosch University — School for Data Science and Computational Thinking
March–April 2026 — using Google Colab
Pandas, Matplotlib, NumPy, Scikit-learn, Seaborn, Statsmodels
Completed April 2026. ✓

QGIS — Basic and Advanced
Ecospatial Lab — spatial data, maps, geographic analysis
Completed May 2026. ✓ (still need a lot of practice and hands-on)

Mathematics
Self-study — calculus, linear algebra, optimisation
This is the hard one. But it's necessary for PhD-level econometrics,
and I'm working on it.

Something I noticed: I started the Python workshop already having built a working tool. The workshop gave me proper vocabulary and structure for things I was doing instinctively. I think sometimes that's the right way to learn — do it first, understand it properly second.

Location:     Rome, Italy 
Role:         Research Intern, Office of Development Effectiveness (ODE), IFAD (UN) — until July 2026
Main work:    Impact assessment taxonomy, gender barriers assessment (ISA), cost-tab analysis
Learning:     Python ✓, QGIS ✓, mathematics (ongoing), web scraping, sensitivity analysis
Presenting:   RES Annual Conference 2026, Speed Session on Development, 6 July, Newcastle
Applying to:  Pre-doctoral and RA roles starting August/September 2026
End goal:     Fully funded PhD in development economics, 2027

If you work on gender economics, household data, development policy, or anything in that space — I'd genuinely love to talk. I'm also always happy to hear from people navigating the MSc to PhD path.

I respond to every message.

📬 porwal.purnima18@gmail.com 🔗 linkedin.com/in/purnimaporwal 🌐 purnima-porwal-dtduwf8.gamma.site