Start with evidence
I listen to users, study the workflow, and separate the visible symptom from the constraint worth solving.
My story
From writing code to shaping AI products and the systems behind them.
Five chapters explain why I look beyond features to users, evidence, model behavior, and operating context.
Developer
I learned what happens after the code works.
Writing software taught me how products are made. Watching some become useful while others went untouched taught me that technology succeeds only when it fits the user, the operating environment, the economics, and the consequences of being wrong.
2017–2021 · SochWare
I learned product by building a company.
I co-founded SochWare in Nepal, helped grow the team from four to more than forty-five people, and contributed to 45+ products. Building across agriculture, healthcare, education, and civic services trained me to enter unfamiliar systems and learn quickly.
2018 · E-Agrovet
Accessible AI became real—not theoretical.
I led an AI product that helped farmers identify crop disease from a smartphone photo and reach expert support. It served hundreds of farmers and won the Microsoft Imagine Cup 2018 World Artificial Intelligence Award.
2022–2024 · Product systems
I moved from shipping products to rebuilding how teams ship.
At CloudFactory, Yarsa Tech, and YHH, I worked across AI data operations, hardware, commerce, and organizational change—connecting discovery, roadmaps, delivery, economics, and durable ownership.
Now · Applied and responsible AI
Today, I build where model behavior meets human need.
My recent work includes accessible audio AI, a research agent with human review, cybersecurity learning for 1,500+ government employees, and a 1,400-response study of LLM sycophancy and truthfulness.
The short version ends here. The full story includes the decisions and lessons behind the work.
Read the full story