Aarav Sharma / aaravfractal.com
Proof of shipping · log 001

I build the layer that catches what breaks in payments.

BTech CSE, second year. I ship production-grade recovery systems and agentic Web3 tooling — Python/FastAPI on the backend, real rails and real chains at the edge. Everything below has a live link, a deployed address, or a verifiable record. Nothing below is a tutorial clone.

Currently shipping Recoup — Track 03, Razorpay AI Buildathon
01 — Selected work

Four systems, each with proof attached

This section is an append-only log. Each entry carries the hash of the one before it, so the chain is only valid if nothing above it changed. Press Verify log in the header to recompute it in your browser.

Recoup — governance layer for failed agentic payments

When an AI agent's payment fails, someone has to decide whether to retry, block, or escalate — and be able to prove why later. Recoup is that decision layer.

Problem
Agentic payment stacks can retry money movement autonomously. Retry logic without a policy gate is an unbounded-spend bug waiting to happen, and without an audit trail nobody can attribute liability after the fact.
Constraints
Solo build, 7 days, hackathon submission. No mocks on the core path — real Razorpay test-mode Orders, payment and capture.
Architecture
Python 3.11 + FastAPI + Pydantic. Deterministic policy gate in front of every money action. Append-only SQLite ledger with SHA-256 hash chaining. Exactly one LLM call, used for diagnosis only and never near execution.
What I owned
End to end — failure taxonomy, policy rules, ledger schema, hash-chain verification, dashboard, demo.
Honest metrics
The dashboard shows attempted, recovered, blocked and escalated separately, plus false-positive and false-negative rates. No 100% recovery claim — the point is that the blocked ones were blocked on purpose.
prev ·sha256computing… · unverified

ChainGuard — scam detection for crypto group chats

A Telegram bot that reads links and contract addresses posted in crypto groups and tells the group, in the thread, whether they're known bad.

Problem
Scam links and drainer contracts spread fastest in the seconds before anyone checks them. By the time a warning is posted manually, people have already clicked.
Architecture
Python (python-telegram-bot) for detection, Google Safe Browsing plus a local blacklist for links, a persistent JSON registry so a scam seen once is re-flagged forever, and a Solidity ScamRegistry contract on Polygon mainnet — verified — written to via web3.py so reports are public and shared.
What I owned
Whole thing, solo. Started as a coursework requirement that only needed a two-page document; the working bot was the part I actually wanted to build.
What went wrong
I leaked a private key into the repo during the mainnet deploy. I caught it, moved funds off the wallet immediately, untracked the .env, and now treat that wallet as permanently burned. Secrets handling on every project since starts from that lesson.
prev ·sha256computing… · unverified

PayParity — confidential salary benchmarking on FHE

Compare your salary against the market without anyone — including the contract — ever seeing your number.

Problem
Salary benchmarking is useful precisely because it's comparative, and impossible to do safely because it's personal. Plaintext on-chain is a non-starter.
Architecture
Fully homomorphic encryption on Zama's stack — comparisons run on encrypted values, so aggregate answers come back without the inputs ever being decrypted. Deployed to Sepolia at 0x063D78A28F677B7D5Da59f8650F94EB739E57175.
What I owned
Solo build and submission to the Zama Developer Program, Season 3 Builder Track.
Why it matters
It's the clearest proof I can point to that I can work at the cryptography layer, not just call an SDK.
prev ·sha256computing… · unverified

Raksha Sutra — digital safety for Indian elders

An AI assistant that helps older users in India recognise the scam before they act on it, in language they actually use.

Problem
Elders are the highest-value target for UPI and OTP fraud in India and the least served by security tooling, which is written for people who already know the vocabulary.
Constraints
Built for a Design Thinking course with Madhav Nauriyal — which meant the interface had to survive testing with actual non-technical users, not just work.
What I owned
Build and deployment; shipped live on Railway rather than left as a prototype.
prev ·sha256computing… · unverified
02 — How I think

Three rules I build by

Taken from things that actually went wrong, not from a blog post.

A system that can move money needs a gate it cannot argue with

LLMs diagnose. Deterministic rules execute. The moment those two swap places you have an unbounded-spend bug, and no amount of prompt engineering fixes it.

If you can't prove it afterwards, it didn't happen

Append-only logs, hash chains, named liability on every event. Recovery is worthless if nobody can reconstruct who decided what.

Report the failures at the same size as the wins

I show blocked and escalated counts next to recovered ones. A dashboard that only reports success is a dashboard nobody should trust.

03 — Distribution & competition

I can build it and I can get people to look at it

Most engineers have one of these. The DevRel case is that I've been doing both since I was sixteen.

500+

Deals closed running Crypto Japans

Web3 marketing agency I've run since 2021 — KOL campaigns and long-standing work with Binance, BC Game and BetFury. Client-facing, deadline-facing, revenue-facing since school.

1st / 800+

Zerodha × FIC national startup challenge

Also 1st at Dehradun Chapters 2026 (E-Cell IIT Bombay / TBI-GEU) with VitaSync AI, and 3rd in AlgoVision at COMET'26, IIT Roorkee.

Knowledge & Research Head, Finance & Investment Cell

GEHU, 2026–27 term. I publish institutional-style research notes — the first covered the NSE F&O market close extension.

StayingAhead — writing and video

My own media brand: newsletter, YouTube and X. Technical explainers for a markets and Web3 audience, written and edited by me.

04 — Stack

What I actually use

Listed only where I've shipped something with it.

Backend
Python 3.11FastAPIPydanticSQLiteREST APIs
On-chain
Solidityweb3.pyPolygonSepoliaZama FHEEtherscan verification
Frontend
HTML / CSSJavaScriptReactDesign systems
Payments & infra
Razorpay Orders / CaptureRailwayVercelGit
AI tooling
Claude / Claude CodeCursorLLM-in-the-loop design
05 — About

Short version

I'm Aarav Sharma, from Ayodhya, currently in my second year of BTech CSE at Graphic Era Hill University in Dehradun. I started in Web3 at sixteen by building an audience and selling campaigns, which taught me distribution before it taught me code. The engineering came after, and it stuck.

What I'm aiming at now is fintech product engineering — payment infrastructure, recovery and reconciliation, the systems that decide what happens when the happy path fails. That's why Recoup exists, and it's why the projects above look less like demos and more like small pieces of infrastructure.

Outside of that: markets research for my college's Finance & Investment Cell, a lot of hackathons, and a media brand I run myself.

06 — Contact

Open to software, DevRel and Web3 builder roles

Internships, contract work, or a hackathon team. Fastest way to reach me is email — I reply the same day.

aarav@aaravfractal.com