How an Agriculture Student in Nepal Can Build a Research Portfolio Without Grants or an MSc

The short answer

An agriculture student in Nepal can build a credible research portfolio without a grant, a master's degree or institutional backing by stacking four low-cost kinds of evidence: a properly designed field trial (usually during the internship), co-authored peer-reviewed papers, open and reproducible geospatial or statistical work, and public documentation that lets a stranger verify all of it. Each piece is small. Together they show a supervisor that you can design, analyse, write and finish.

This article uses one real path as the worked example: that of Bishal Neupane, a 2025 BSc Agriculture graduate of Agriculture and Forestry University (AFU), Rampur, who by 2026 had nine peer-reviewed articles and is now seeking an MSc research assistantship. The ideas are drawn from two of his LinkedIn posts, in which he describes himself as a student in Nepal with "no MS, no research grant, and no institutional backing."

Why a specific problem beats a general skill list

In the first post, Neupane's main lesson is that the people he contacted responded to a specific research problem, not to a degree or a list of skills. His example question was why global crop models struggle with Himalayan farming systems. A line such as "I know GIS, R and Python" is common in applications; a line that names a problem in a place, and shows work already done on it, is not.

The practical consequence is that every portfolio item should connect to a question. A land-cover map is a skill demonstration. A land-cover map that asks how seasonal water presence changes around a lake, and says what the analysis cannot yet tell you, is research.

Step 1: turn your internship into a real experiment

Most Nepali BSc Agriculture programmes include a field internship. It is the cheapest research opportunity you will ever get: land, inputs and supervision are already provided. Treat it as a trial, not a placement.

Neupane's internship at the Prime Minister Agriculture Modernization Project (PMAMP) in Morang, from January to July 2025, is a good template. He designed and ran a randomised complete block design (RCBD) experiment on okra, testing three concentrations of the plant hormones NAA and GA₃ on more than 50 plants, collected over 200 morphological and yield observations, and analysed them in R with ANOVA and LSD. GA₃ at 100 ppm was the most effective treatment. That single experiment became a first-author manuscript accepted in June 2026 by the Turkish Journal of Agriculture – Food Science and Technology (in copyediting at the time of writing; see the summary of the okra study).

What made it publishable was not scale but design: replicated blocks, a control, a clear response variable, and a statistical analysis a reviewer could check. If you want to do the same, the RCBD analysis guide in R covers the full workflow.

Step 2: co-author with your peers, and do real work on each paper

Final-year students across Nepal run trials and surveys for their own internships and theses. Many of those datasets never become papers because the student lacks time, statistical help or writing experience. Offering to analyse, write, or handle submission and revision is a legitimate way to contribute, and it builds a record quickly.

Neupane's 2026 record shows what this looks like. Of his nine published articles, he is first author on three and corresponding author on all nine; on the others he is second or third author, with the student who ran the trial or survey as first author.

Three further manuscripts are accepted and in press. The full list, with each paper's question and main finding, is on the publications page.

Two rules keep this honest. Authorship order should reflect who did the work, and the corresponding author must be able to answer for every number in the paper. Choose journals with DOIs and a clear peer-review process, and read the final proofs line by line.

Step 3: do open, reproducible geospatial and data work

Field trials show you can run an experiment. Precision agriculture supervisors also want to see that you can handle spatial and temporal data. Free tools make this possible from a laptop in Pokhara: Google Earth Engine for satellite imagery, R and Python for analysis, QGIS and OpenStreetMap for mapping.

• Satellite analysis. Neupane learned Google Earth Engine largely on his own and built a Kaski project using dry- and wet-season Sentinel-2 composites, five vegetation indices (NDVI, NDWI, SAVI, GCI and PSRI) and exploratory K-Means clustering into six land-cover classes, explicitly noting that field validation is needed before operational use. See the Kaski land-cover walkthrough.
• Time series. A multi-temporal NDWI analysis of Pokhara Lekhnath tracked seasonal pulses of water presence, described on the methods page.
• Volunteer mapping. As an Open Mapping Guru with Open Mapping Hub Asia Pacific since 2024, he has validated more than 5,000 geographic data points and corrected over 300 across 15+ regional datasets, working with eight international researchers.
• Code. Applied machine-learning work, such as a crop disease detection app using TensorFlow and MobileNetV2, is public on GitHub.

Honest caveats, like the field-validation note on the Kaski project, make this work more credible to an academic reader, not less.

Step 4: document everything in public

A portfolio nobody can find does not work as a portfolio. Short, accurate posts about what you did, what went wrong and what you learned give supervisors a way to check your claims before they reply. Neupane uses LinkedIn for research notes and a personal academic website that links every paper to its DOI, every project to its code or post, and his CV to a downloadable PDF.

Step 5: reach out directly, and keep going

In the second post, Neupane looks back a year to when he was counting tillers in a rice field, and describes moving from there to having eight manuscripts under review, analysing farmland with Sentinel-2 imagery and building models to predict crop yield. He says he contacted more than 100 professors worldwide, and argues that agriculture graduates should not wait for offers to arrive, because "opportunities don't find you in a waiting room."

Direct outreach works best when the email is short, names the professor's specific project, states one question you would like to work on, and links to evidence. A portfolio built through steps 1 to 4 is what makes that link worth clicking.

What a portfolio does not replace

None of this substitutes for formal graduate training. A portfolio built alone can show initiative, method and output; it cannot provide the depth of supervision, theory and peer critique that an MSc gives. Its purpose is to show a supervisor that the investment in you is likely to pay off. Neupane's own next step is exactly that: an MSc research assistantship in precision agriculture and geospatial science, described on his research-fit page.

Students who want to compare notes, and supervisors looking for a motivated graduate researcher, are welcome to reach out through the contact page.