Backend leaning software dev who likes clean APIs, measurable impact, and practical AI. I build with Python & Java, ship services with Flask / FastAPI / Spring Boot, and exploring RAG (Retrieval-Augmented Generation) with LangChain.
- RAG for end-to-end HR automation - PRD -> eval-driven prototype with retriever, prompt augmentation, and RAG eval (faithfulness / answer relevance).
Repo: (create)piyushsatti/rag-hr-automation->/ingest,/retriever,/orchestrator,/eval(RAGAS).
-
Nonagon — partial ETL + API app (Flask, FastAPI, SQLite, Discord SDK).
Repo • API scaffolding, auth, CRUD, feature toggles. -
risk-emulated — Java strategy-game engine using MVC, graph maps, and Strategy/State/Command patterns.
Repo • testable AI agents + map builder. -
MMAPF — Min-Max Average Pooling Filter for impulse noise removal (published in IEEE SPL). Repo • DOI.
-
class-to-sql — micro-ORM style Python utility to work with classes via SQL-ish ops. Repo • packaging/tests WIP.
Languages: Python, Java, JavaScript/TypeScript, SQL
Frameworks: Flask, FastAPI, Spring Boot, Express
Data/AI: RAG, LangChain, PyTorch, scikit-learn, Pandas, NumPy
Infra/Tools: Docker, GitHub Actions, Postman, Pytest, JUnit
- Ship in small, measurable increments; keep quality visible (tests, clear PRs).
- Treat AI as default tooling for scaffolds/search/log triage—own the design & correctness.
- Prefer simple, well-named modules over premature abstractions.
Website: https://piyushsatti.github.io • LinkedIn: https://www.linkedin.com/in/piyush-satti/ • Email: piyushsatti@gmail.com