Ayush Rai
Software Engineer building AI systems, focused on automation and developer tools.
Personal knowledge system: what I ship, what broke, and what I would do differently. Evidence lives in the pages below. Start wherever curiosity points.
ayushrai0211@gmail.com (reply within 24 hrs) · +91 7440567944 · GitHub · LinkedIn · X · Linktree
Current Focus
- AI coding-agent evaluation ; Harbor and Terminus task systems, Dockerized oracles, review CLIs
- Host security auditing ; read-only DFIR-style inventory with MITRE-mapped findings
- Interpretable business analytics ; constrained P&L scoring instead of proxy-target ML
- RAG and agentic LLM applications ; routing, retrieval, and local-first experiment systems
- Compliance and audit automation ; including deterministic, non-generative designs
Latest Work
- RevenueRoles ; technical revenue job intelligence and job board
- RLHF-tasks ; coding-agent evaluation toolchain (Harbor / Terminus)
- security-audit ; read-only cross-platform host audit toolkit
- amex ; deterministic P&L scoring for 500k cardmembers
- OpenAI-RLHF ; coding-eval task archive (not an RLHF trainer)
- IAS-system ; OCR → SOP → score audit pipeline
Recently Updated
- Currently Reading ; What I'm reading, learning, and building from.
- Knowledge base ; Identity inventory and system status refreshed.
- Currently learning ; Active study and practice areas from public evidence.
- GitHub ingest ; Public repositories indexed for projects and latest work.
- Projects index ; Canonical project entities under knowledge/projects.