Working on LLM and full-stack problems at Doozie Software Solutions.
Software developer · Bengaluru
Ashwin K M
I build web applications, work with AI tools, and write down what I learn from real projects.
Current signal
What I can point to right now.
A small set of public work and learning direction. If something is not shipped, started, or clearly next, it does not belong here.
AI Risk Evaluation Workbench
A lightweight evaluation workbench for comparing open-source and frontier AI assistants before enterprise deployment.
View repositoryGitHub issue resolver plugin
An OpenClaw plugin for fetching and analyzing GitHub issues, built from the hackathon work I wrote about.
View repositoryInference and ML fundamentals
Tracked on the now page: Inference Engineering and DeepLearning.AI's Fine-tuning and Reinforcement Learning for LLMs.
See current focusKeep the site close to the actual work.
Selected writing
Posts from the archive.
I won my first hackathon by building something I barely understood
4 hours, no plan, and an autonomous GitHub issue resolver that almost didn't work. Notes from the HSRFC x OpenClaw Builders event in Bengaluru.
Recent posts
The latest things I've published here.
- Essay 6 min read
Karpathy's 20/80 Rule: Why the Best Engineers Will Write Less Code, Not More
Andrej Karpathy operates at a 20/80 human-to-agent ratio. The bottleneck in engineering has moved — from writing code to orchestrating intelligence.
- Essay 3 min read
I won my first hackathon by building something I barely understood
4 hours, no plan, and an autonomous GitHub issue resolver that almost didn't work. Notes from the HSRFC x OpenClaw Builders event in Bengaluru.