CloudSlash

Local-first AWS forensic engine. Finds waste via dependency graph analysis and enables safe remediation with Terraform state restoration.

Inspiration

I created CloudSlash as a personal project to decompress from medical school. My med school training introduced me to Bayesian algorithms, and I wanted a creative outlet to apply probabilistic thinking to real problems. I’d also done CS50x back in 9th grade and was familiar with algorithmic ideas like the bin packing problem — those concepts inspired how I approached cloud resource analysis. I built the project myself and used AI tools where they helped speed up development.

What it does

CloudSlash analyzes AWS environments locally to find waste and risky configurations by building dependency graphs and applying algorithmic heuristics. It:

Detects unused or wasteful cloud resources and surfaces them with context.
Prioritizes findings using probabilistic and optimization techniques (Bayesian reasoning, packing/optimization intuition).
Produces safe remediation guidance and supports restoring Terraform state to avoid destructive changes.
Is local-first so sensitive cloud data stays under the user’s control.

How I built it

Implemented the core engine in Go, with a few small helper shell scripts.
Used Bayesian approaches I learned in med school combined with optimization ideas (inspired by bin packing) to rank and consolidate waste candidates.
Applied AI-assisted tooling for parts of the workflow where it accelerated analysis and iteration.
Iteratively developed features and safety checks around Terraform state remediation to make recovery straightforward and non-destructive.

Challenges I ran into

The main hurdle was adoption — getting people to try the tool and provide real-world feedback.
To solve that, I reached out on Reddit and other places, asked for help and feedback, and implemented the changes users requested.
Iterating quickly on community input helped the project become more useful and practical.

Accomplishments that I'm proud of

I built CloudSlash end-to-end on my own, from the algorithmic core to the user-facing workflows.
The project successfully combines principled algorithmic thinking with practical cloud forensics and safe Terraform remediation.
I received positive community feedback — people liked the idea and found it useful.

What I learned

Concepts from other fields (like Bayesian reasoning from med school) can translate well to cloud engineering problems.
User feedback is invaluable — real users expose the most important improvements to make.
Combining human-designed algorithms with AI assistance can accelerate progress while keeping control and safety intact.
Designing for local-first usage and careful Terraform state handling increases real-world trust and adoption.