AI Safety Nepal

(website: https://www.aisafetynepal.org/)

Project Summary

AI safety fellowship to upskill local talent, build a pipeline of people who understand AI safety deeply enough to contribute to research, advise on policy, and coordinate when AI governance decisions are being made.

What this project is

Following the interest of participants from our inaugural AI Safety Nepal kick-off event (LinkedIn), we want to provide a stepping stone for students and professionals to engage in the field of AI safety. This will be Nepal's first dedicated AI safety fellowship, in-person and discussion-based, and will use the AI Safety Atlas curriculum as its core material over five weeks. The group discussions will be in groups of five and will be facilitated by experienced researchers and engineers in AI and AI safety. We also plan to have a few guest lectures from AI safety researchers and governance practitioners.

In the final weeks, participants develop either a technical AI safety research or governance/policy project where we plan to take the 3-5 most promising projects forward with mentorship and compute support after the program.

Theory of Change

Through the fellowship, we want to encourage and upskill talented folks in AI safety. Nepal has talented students and professionals, but AI safety as a field is not known to many. We want to upskill local talent so they understand AI safety better, understand pathways for further contribution, and contribute to global AI safety efforts through independent research, participating in fellowships like SPAR, MARS, MATS, etc., and contributing to policy and local governance efforts.

Over 70 people applied for our kick-off event within a week, among whom we selected 50, and more than 85% expressed interest in a longer-commitment fellowship. This gave us confidence that there is interest in this kind of program. With this fellowship, our aim is:

• Discussion-based learning with participants from different disciplines.
• Become an on-ramp for students and professionals to transition into AI safety.
• Identify high-potential participants and continue mentoring 3–5 promising research or governance projects beyond the fellowship.

What this fund will be used for

This fund will be used for our five-week fellowship ($5527), followed by three-month project mentorship ($5500). Here is the budget breakdown.

Our Team

Yunika Bajracharya, Founder and Executive Lead
Current research: Quantifying AI Alignment Failure Risks
Presented the research and co-ran expert elicitation workshop at LSAIR 2026
Research Fellow at AI Safety Saarland Interdisciplinary Research Incubator
BlueDot AGI Strategy Course & Technical AI Safety Course
Prev. Machine Learning Engineer at REG-1
Website, LinkedIn, GitHub

Anju Chhetri, Co-Operations Lead
Role: Research Assistant at BBMLL lab (>2 years)
Primary Affiliation: NepAl Applied Mathematics and Informatics Institute for research (>3 years)
Research Area: Machine learning robustness (Out-of-distribution detection), Interpretability
HAAISS (2026), EEMLL(2025) Scholar
Website, LinkedIn, GitHub, Google Scholar

Rujal Acharya, Co-Operations Lead
Research Engineer at  Accelerated Komputing
Current research: Precision in Computer Numbers
Previous research: AI for Dental Crown Generation, Physics AI
BlueDot AGI Strategy Course & Technical AI Safety Course
Google Scholar LinkedIn

Tripti Sharma, Communications Lead
Communications Lead at NepAl Applied Mathematics and Informatics Institute for research
Current research: Subliminal Learning (blog)
Research Assistant at AI and Society Lab
Program Lead, 7th Annual Nepal AI School 
BlueDot Technical AI Safety Course and AI Safety Project Sprint
Fellow at Young Women's Political Leadership Course
LinkedIn, Substack

Pramesh Shrestha, Public Relations Lead
AI Engineer at REG-1
Co-founder of Edavor
Organizer of Github Field day Nepal
Prev. Researcher at KTM Geo Lab - Duke University x Pulchowk Campus
LinkedIn, Github