Work
Proof of work: what I have built, shipped, researched, and contributed to.
Each entry links to the page that holds the evidence, here or on ayushrai.site, instead of copying it. Labels say how far each piece got.
Proof of work by role
Each role has its own page on my main site, ayushrai.site. Links marked ↗ leave this site.
-
AI Benchmarking (on ayushrai.site)
Evaluation systems that turn model behaviour into measurable, reproducible evidence.
-
AI systems that combine reasoning, tools, state, and verification to run multi-step tasks with control.
-
Forward Deployed Engineer (FDE) (on ayushrai.site)
From ambiguous problem statements to working software for operational problems. The page marks its demonstration as in progress.
-
AI Research Engineering (on ayushrai.site)
Reproducible experiments that connect research questions, implementations, and evaluation.
-
Founding Engineer (on ayushrai.site)
Zero-to-one product work across product thinking, full-stack engineering, AI systems, and infrastructure.
-
DevOps Engineering (on ayushrai.site)
Automating the path from code to reliable production: reproducible delivery, safe releases, predictable recovery.
-
Machine Learning Engineering (on ayushrai.site)
Machine learning systems beyond the notebook: data pipelines, training, evaluation, inference, and reliability.
-
Product Management (on ayushrai.site)
Turning ambiguous problems into products people can use, worked through on RevenueRoles.
-
GTM Engineering (on ayushrai.site)
Data and decision systems for revenue work: data pipelines, entity resolution, explainable scoring, and approval workflows.
Start here
- RevenueRoles ; Products and applications
- Adaptive RAG ; AI and machine learning
- GTM engineering ; GTM engineering and business systems
- Public policy work in 2025 ; Public policy and civic work
Products and applications
Things people can use, or will be able to.
-
RevenueRoles · Live ·
Job discovery for technical revenue roles such as GTM engineer, RevOps engineer, and forward deployed engineer, classified by the work a posting describes rather than by its title.
-
Shiksha-Mitra · Live ·
A collaborative learning platform where peers share resources, acknowledge progress, and support each other's career and education journeys.
-
The Open Framework · Live · · on ayushrai.site
An early digital public infrastructure concept for civil-society coordination, published as a front-end prototype rather than as live national infrastructure.
-
AGENYRA · Live · · on ayushrai.site
A distribution layer for AI products, starting with structured representation and builder onboarding, with discovery and placement still at earlier stages.
-
MedVoice · Public code · · on ayushrai.site
A voice-assisted order and billing workflow for pharmaceutical wholesalers, where speech proposes a line and deterministic rules decide identity, tax, and posting.
AI and machine learning
LLM applications, retrieval, and evaluation work.
-
Adaptive RAG · Public code · · on github.com
Agentic retrieval-augmented generation with LangGraph: query rewriting, relevance grading, web-search fallback, and routing between retrieval, search, and plain reasoning.
-
MailU · Public code · · on github.com
Local email intelligence: Gmail API sync, zero-shot categorization, semantic search, and RAG chat over your own mail on a local model.
-
OpenAI-RLHF · Public code · · on github.com
An archive of coding-evaluation tasks with a task generator, test runner, and validation tooling. It is not an RLHF trainer.
-
RLHF-tasks · Public code · · on github.com
Coding-agent evaluation toolchain: Harbor and Terminus task systems, Dockerized oracles, and review CLIs.
GTM engineering and business systems
Data and decision systems for revenue work.
-
GTM engineering · Live · · on ayushrai.site
Data and decision systems for revenue work: job-market data pipelines, entity resolution, explainable scoring, and human-approval workflows. The page covers RevenueRoles, the Revenue Intelligence Engine, the Relationship Intelligence Engine, SpiderWeb, and MailU, and separates what I have built from what I am still learning.
Software engineering and infrastructure
Audit, compliance, and security tooling with deterministic behaviour.
-
security-audit · Public code · · on github.com
Read-only, cross-platform host audit toolkit with DFIR-style inventory and MITRE-mapped findings.
-
SoPia · Public code · · on github.com
Deterministic SOP compliance auditor.
-
IAS-system · Public code · · on github.com
Judgment and case-file audit pipeline: OCR, SOP checks, then scoring.
Research, analysis, and experimentation
Analysis where the method matters as much as the result.
-
amex · Public code · · on github.com
Interpretable business analytics: constrained P&L scoring instead of proxy-target machine learning.
Public policy and civic work
Policy research, drafting, and scheme review.
-
Public policy work in 2025 · Completed ·
GCTP 2025, about five drafted policy proposals on geopolitical contexts in India and city development, a review of government schemes, and a three-day public policy challenge.
Ongoing personal projects
Long-running work that is not public yet.
-
A book · In progress
I am writing a book. The subject and timing will be shared when it is ready.
-
A long-running application · In progress
An application I have been building over roughly six to seven years. It is in development and not public yet.
How to read the labels
- Live
- A public URL serves it today.
- Public code
- Code is in a public repository. No hosted deployment is claimed.
- Completed
- Finished. The page says what was produced and what was not.
- In progress
- Not finished and not public yet.
Metrics appear only where the linked page can back them. A public repository means the code is public, not that the product is finished.