// Md Athar Sharif · IIIT Naya Raipur · CSE '28

I build systems that have to be right, not just running.

Backend & distributed-systems engineer first — consensus, ZK-verified contracts, production apps. Full-stack when it counts, Kotlin everywhere, Linux tools for fun. Verified with simulators, benchmarks, and 43 real releases — everything here has commits behind it.

github/sharifmdatharsee the work ↓
96repos
4live deployments
50+on-chain users
9.05CGPA

## selected work

ChainChit

live ↗

On-chain chit fund (rotating savings) platform on Stellar. 5 Soroban smart contracts in ~3.9K lines of Rust: group lifecycle, sealed bidding, identity, reputation, dispute resolution.

  • 53 testnet users · real on-chain activity
  • Black Belt, Stellar Journey to Mastery ($150 USDC)
  • CI on every push · factory-pattern deployment

rust · soroban · next.js · typescript

AegisProcure

live ↗

Trustless sealed-bid reverse auctions for B2B/GovTech procurement on the Midnight blockchain. Bids stay confidential end-to-end — the chain verifies validity without ever seeing a number.

  • 4 ZK circuits: create · commit · reveal · finalize
  • Compact smart contracts, live on Midnight Preprod
  • Vitest contract tests gated in CI

zk-proofs · compact · midnight · next.js

IIITNR Inventory System

live ↗

Cross-platform inventory system — Android, iOS, desktop, web from one shared Kotlin codebase. Officially adopted by IIIT Naya Raipur to manage institute-wide inventory.

  • 43 releases through automated build-and-sign CI
  • Offline-first (SQLDelight) · JWT + role-based access
  • Nightly DB backups · concurrency-safe stock updates

kotlin multiplatform · fastify · postgres

TrustVault

live ↗

Decentralized escrow on Stellar with built-in arbitration and fee sponsorship — trustless buyer/seller transactions with contract-level timeout enforcement.

  • Level 5, Stellar Journey to Mastery ($100 USDC)
  • Escrow + arbitration contracts · CI/CD

soroban · react · typescript

MARL Fleet Rebalancing (research)

Measurement study: how do frozen RL fleet-rebalancing policies degrade when deployment demand differs from training demand? 62 NYC zones, real ride-hail demand, IPPO vs MAPPO vs tuned LP baselines.

  • Parameterized demand-shift scenario matrix (S0–S6)
  • Reproducible train/eval pipelines + test suite

python · pettingzoo · reinforcement learning

Aximo

Voice-controlled AI shell agent: natural language → validated system commands via Gemini, with security checks against malicious execution and graceful API-failure recovery.

  • Cross-platform (Linux, macOS, Windows)
  • Documented REST API (Swagger)

python · fastapi · llm

## rust tooling & open source

Dynamic IslandQML media/notification widget for Omarchy (Hyprland) — published & verified in the official plugin marketplaceTuisharklive packet-sniffer TUI — pcap capture, TCP dissection, GeoIP, ratatuiudedhardware-diagnostics CLI for used devices — CPU, disk, battery, wifi/rfkill, audio checksTarSmithinteractive installer for tar-archive Linux distributionsnm-notifylightweight Wi-Fi roaming notifier daemon for NetworkManagermini-redisfrom-scratch Redis-compatible server over async TCP (learning project)

## achievements

## today i learned

Sep 22, 2026

discovered raft today

it's a leader-based concensus algo for distributed systems to ensure fault tolerance and consistency by managing a replicated bug.

all 1 notes → · rss

## stack

languagesRustKotlinPythonTypeScriptC/C++SQL
backend & systemsSoroban / StellarMidnight / CompactZK proofsRaft (in progress)PostgreSQL / Drizzle
full-stack & appsNext.js / ReactKotlin MultiplatformFastifyFastAPI
linux & toolsRust CLIsGitHub Actions CIDockerLinux / NixHyprland (daily driver)
ai / mlRL — PettingZoo / IPPO·MAPPOagent-assisted engineering

// currently

Building a from-scratch Raft consensus implementation with a deterministic simulation harness — 0-violation runs and benchmarks to follow. Open to internships and systems/ZK engineering collaborations.

email meLinkedIn ↗GitHub ↗