AI dev enablement
Arunkumar Ganesan / @akg268
Backend systems, developer tools, and production feedback loops.
I use GitHub as a working notebook for practical software: the Thule AI Dev Skills project, prompt-preflight tooling, RAG and MCP experiments, Java/Spring systems, gRPC services, testing examples, and the kind of production learning loop that makes the next release safer.
0Public repos
LoadingSource projects
0Open project issues
0Community signals
Developer tooling
Prompt-preflight
Local hooks for catching vague agent prompts before they waste time or produce the wrong kind of work.Applied experiments
LangChain, RAG, and MCP
Small Python repos for testing retrieval, model context, tool boundaries, and practical integration patterns.Backend systems
Spring, gRPC, Kafka, OAuth
A long-running Java/Spring trail around services, tracing, messaging, authentication, and platform reliability.Quality loop
Everyone tests in production
The point is not chaos. It is observability, small rollouts, fast rollback, and turning real behavior into better tests.Featured project
Thule AI Dev Skills
Thule AI Dev Skills is about making AI-assisted development feel like a real engineering practice, not a shortcut. The project focuses on the habits that make agents useful in production code: clear task framing, grounded context, testable changes, useful reviews, and honest handoffs.
Skill map
Break AI-assisted development into teachable habits: framing the task, gathering repo context, asking for the right evidence, and knowing when to stop.
Engineering loop
Pair agent speed with normal software discipline: tests, code review, small commits, clean handoff notes, and production-minded risk checks.
Practice system
Turn everyday development work into repeatable exercises, examples, and rubrics that help engineers build judgment instead of just using a tool.
Projects
From prompt guardrails to Spring-era systems craft.
The GitHub trail moves from Java backend fundamentals into newer tooling and model-context experiments. Thule AI Dev Skills connects that work to engineering enablement, while recent repositories lean into prompt quality, RAG, LangChain, MCP, and Spring application work. Older repos keep the Spring, Kafka, gateway, OAuth, and testing foundation visible.
Project signal
Fetching public repositories from @akg268.
GitHub APILive
Project signal
Fetching public repositories from @akg268.
GitHub APILive
Project signal
Fetching public repositories from @akg268.
GitHub APILive
Project signal
Fetching public repositories from @akg268.
GitHub APILive
Everyone tests in production
The craft is making production feedback intentional.
Production is where traffic, timing, data shape, and human behavior finally meet. The right engineering stance is not pretending that this never happens. It is designing the loop so every surprise has a guardrail and every signal becomes better software.
Observe
Trace the real path through logs, metrics, and user-visible signals.
Gate
Use flags, canaries, and rollbacks so experiments have edges.
Learn
Compare expectations against production behavior, then keep the useful surprises.
Patch
Move what you learned into tests, hooks, runbooks, and safer defaults.
Contribution search
Find unassigned issues no one has commented on.
Search for open issues across GitHub by theme, then open only the untouched matches: no assignee and zero comments. Try terms around tooling, testing, Spring, observability, documentation, or whatever you want to practice next.
Search GitHub for unassigned issues with no comments.