jayantx.com

Solana engineer. Zero-knowledge and post-quantum cryptography.

From the matching engine
to the STARK proof.

Right now I am at Bonsol, building Sombra, a post-quantum private payments protocol on Solana. I built the wallet end to end and shipped it to the Chrome Web Store, including the native proving daemon it talks to. Kafka, Redis and a Node.js matching engine at one end of my stack, RISC Zero STARK proofs and ML-KEM-768 at the other.

Before Bonsol: a CEX simulator with a heap-based matching engine at 100xDevs, a RabbitMQ ingestion pipeline at Tealfeed where I cut API response times 40 percent, and a team-objectives SaaS I led at GEA.

I write about the cryptography while I am learning it, and ZK from Scratch #1 has 188,000 impressions. Fuin is also mine, an on-chain permissions layer. The Solana Foundation backed it with $10,000.

I am at Bonsol full-time. If you are building on proving systems, private payments, or the infrastructure underneath them, send me the problem.

Separately, I direct AI films. Seedance for generation, Remotion for the post-production overlays and titles.

188K+

impressions on ZK from Scratch #1

$10,000

Solana Foundation grant, to Fuin

42K+

views on the Solana indexer demo

40%

cut in API response times at Tealfeed

SuperDev Fellowship, Solana FoundationSolana India Fellowship 2026YC Startup School India 2026Turbin3 graduateSuperteam IndiaSuper30

What actually shipped.

Protocols, infrastructure, a trading engine, all of it running.

Sombra · Bonsol · 2026 to now

One undecryptable note could lock a user out of their own ledger. The fix had to go inside the proof.

A single UTXO the wallet could not decrypt was enough to brick a private ledger. The fix could not live in the client, because the client is not what the chain trusts, so the guest program now skips undecryptable notes inside the proven statement.

The proof

Client-side RISC Zero STARKs. Hash-based and transparent, generated in the browser extension.

The secret

ML-KEM-768 per UTXO, ChaCha20-Poly1305 payloads, AES-256-GCM for pending jobs.

The wallet

Chrome MV3, multi-account keys, UTXO batching, local Rust daemon or a remote prover.

CEX simulator · 100xDevs · 2025

Building a matching engine taught me more about queues than about markets.

Min and max heaps: O(1) to read best bid and ask, O(log n) to insert and fill. Kafka sits between the engine, the database processors and the WebSocket fan-out, so a slow consumer cannot stall a fill. Liquidation handles isolated margin and leverage.

The engine

Node.js. O(1) to read the best bid and ask, O(log n) to insert and fill.

The plumbing

Kafka between the engine, the DB processors and the WebSocket fan-out.

The risk

Liquidation across isolated margin and leverage.

I teach it as I learn it.

ZK from Scratch started as notes to myself. The first one has 188,000 impressions, so I kept going.

Where I have worked.

Jan 2026 to now

7 mos

Bonsol

Solana Engineer

United States, remote

  • Core engineer on Sombra, a post-quantum private value-transfer protocol
  • Client-side RISC Zero STARK proofs, ML-KEM-768 per UTXO, signature-free ownership
  • Built the wallet end to end and shipped it to the Chrome Web Store, native proving daemon included

Aug 2025 to Dec 2025

4 mos

100xDevs

Software Engineer

Noida, India

  • Order matching engine in Node.js on min and max heaps, O(1) to read the best bid and ask
  • Kafka microservices decoupling the engine from the database processors and the WebSocket fan-out
  • Liquidation across isolated margin and leverage, plus a visual workflow automation tool

Oct 2024 to Aug 2025

10 mos

Tealfeed

Software Development Engineer I

Gurugram, India

  • RabbitMQ pipeline for large CSV and Excel uploads, which ended the server timeouts
  • MongoDB query tuning, 40 percent faster API responses
  • Retry-safe WhatsApp and email notification service

Aug 2023 to Jul 2024

11 mos

Global Entrepreneurship Alliance

Full-stack Developer, part-time

Belgium, remote

  • Led True North, a team-objectives SaaS on Next.js and Prisma
  • OpenAI integration for automated analysis of user data
  • Refactored the frontend into reusable components, 25 percent faster feature delivery

What I build with.

Languages

Rust and TypeScript daily

RustTypeScriptJavaScriptSoliditySQL

ZK and crypto

Primitives I implement, not libraries I call

RISC ZeroSTARKsML-KEM-768ChaCha20-Poly1305AES-GCMFiat-Shamir

Solana

Programs, indexing, oracles, on-chain proofs

AnchorPDAsSPL TokenGeyser gRPCPythOn-chain verification

Backend

The systems half: queues, matching, fan-out

Node.jsKafkaRabbitMQWebSocketsRedisPostgreSQLDocker

Frontend

Including an extension live on the Chrome Web Store

ReactNext.jsChrome MV3Tailwind

Build with me