Founding Engineer · New York, NY
Marlon Ma

Zero-to-one across mobile, payments, on-chain, and AI. Founding engineer from idea stage to $3M pre-seed.

github.com/marlon-maDownload résumé (PDF)
Beyond the resume

AN ENGINEERBEYOND THE RESUME.

I build for a living, but I'm not just an engineer. The same thing that makes me ship on weekends is what makes me pick up a camera, a brush, or a new language for fun.

A few things you won't find on the PDF

Founded a SaaS for photographers

Picked up a camera in college and never put it down. That obsession turned into a SaaS for studio photographers — my first real product, shipped before I ever called myself an engineer.

Calligraphy practitioner

Chinese brush calligraphy since I was a kid. Patience, line weight, knowing when to commit — turns out it's a surprisingly good warm-up for code reviews.

Native Mandarin · fluent English · learning Arabic & un poquito Spanish

Born and raised in China. Trained in the US. Equally at home reading a Beijing engineering blog or pair-programming with a New York founder. Currently learning Arabic, and picking up un poquito Spanish on the side.

Long runs, espresso, sci-fi

Morning runs along the Hudson. Spends more time choosing beans than coding tools. Reads more sci-fi than business books — that's where the good systems thinking lives.

STUDIO PRACTICE

Self-taught artist. Maybe.

Long before I wrote production code, I was teaching myself to see — through a camera, a brush, a darkroom. The studio is where I learned the things engineering can't teach you: patience, taste, knowing when a thing is done. This is one of mine.

The Pomegranate's Skin — A short film. Self-directed.

Same hands that ship the code make the work. Same eye for detail.

By the numbers

10 months. Blank repo.

MO

Blank repo → live consumer app

Commits owned at Hawala

%

Of entire repo

Teams running on my AI platform

Pre-seed — joined idea-stage

s

Median AI support resolution

Founding Engineer · Jul 2025 – Present

HAWALA

Stablecoin neobank for emerging markets — first engineering hire.

I joined at the idea stage. Today the app is live on the App Store and Play Store, the company has raised a $3M pre-seed, and the consumer mobile surface, the custodial card program, the cross-chain settlement, and the internal AI platform six teams query every day all run on systems I built or rewrote.

Mobile from blank repo

Shipped Expo template → App Store + Play Store in 10 months. Own architecture, design system, release engineering, on-call. 2,008 commits over 10 months — ~38% of total repo.

React NativeExpoTypeScriptEAS

Every consumer flow

International onboarding, identity verification, multi-rail money movement (cross-chain send/receive, cards, on/off-ramps, investing, swap), referrals & rewards. End-to-end funnel and performance.

CardsOn/off-rampsReferrals

Cross-chain settlement

Dual-chain non-custodial wallet on Solana + EVM. Sponsored EVM↔Solana USDC bridging, unified history, idempotent settlement. Multi-step bridges collapsed into one user action.

SolanaEVMUSDCPrivy

Backend + data

~300 PostgreSQL migrations and 178 edge functions on Supabase. Integrated 4 financial partners. 80%+ test coverage on edge functions.

DenoPostgreSQLSupabase

Two flagship AI agents

Tier-1 support agent grounded in live partner-rail context, serving 6 languages — replaced the English-only human queue. Engineering debug agent that kicks off from Slack and drafts root-cause PRs.

LLM orchestrationRAGTool use

Engineering bar as team grew

Coding standards, design system, folder-structure ESLint rules, Zod runtime contracts mobile↔backend, migration-first DB writes. Hundreds of PRs reviewed.

ZodESLintMigration-first
INTERNAL AI OS

AI PLATFORM

Not a chatbot. A single-agent multi-role shared-context runtime with memory, tool use, approvals, audit, and evals — grounded in retrieval against a Git-backed llms.txt substrate of ~180 source docs. Used daily by engineering, ops, PM, growth, marketing, and support.

Consumers · 6 teams
EngineeringOpsPMGrowthMarketingSupport
Shared runtime
Single-agent · multi-role
shared-context runtime
memorytoolsapprovalsauditevals
Retrieval substrate
llms.txt substrate
~180 docs · Git-backed
Outcomes · measured
  • Hours → <10 min
    Time-to-RCA
  • 9–38s median
    Tier-1 support, six languages
  • Eval-driven rebuild
    Benchmarked then rebuilt
  • +28% retrieval
    Retrieval substrate
  • Zero auto writes
    Safety by construction
WHAT'S NEXT

AI-native fintech is not a chatbot bolted onto a form.

The product itself is the agent. The user states intent — "send tuition to my daughter at NYU," "settle this invoice in Dubai" — and the system reasons about corridor, compliance, FX, timing, and counterparty trust on their behalf. KYC is a conversation. Risk is continuous. Ops scale with intelligence, not headcount.

01

Ameen

The first agent in the layer.

Support, KYC, declines, transfer issues. One agent doing what hundreds of people would. Nine languages. Every ticket becomes training data.

02

Ask Hawala

Same engine, turned to the user.

Personal finance in plain English. How much did I spend on coffee. Send mom a thousand dollars for Eid. Should I move savings into stablecoins.

03

Agentic Economy

The wallet for the next trillion users — who won't be human.

Agents pay each other in stablecoins, in micropayments, with programmable limits. We build this for emerging markets while the West builds it for the West.

BEFORE HAWALA

EVERYTHING BEFORE THIS

Founder · Mar 2022 – Jun 2024

Photography Studio SaaS

Ran a 4-person team on a full-stack booking + portfolio platform for photographers. 60+ paying clients, $15k+ revenue. Payments, S3 CDN, a CRM that cut studio overhead 50%.

FounderFull-stack$15k+ rev

Undergrad Researcher · Northeastern AI Lab · Sep 2021 – Jun 2024

Industrial defect detection

Designed YOLOv5-G (high-resolution P2 head, RepGFPN multi-scale fusion, Kmeans++ anchor clustering) — +8.3% small-object mAP. Expanded a 741-image set to 2,124 via Mosaic + geometric augmentations. Shipped end-to-end to a glass manufacturer's QC line: detection precision 60% → 82%, manual labor down 60%.

YOLOv5-GRepGFPN60% → 82% precision

Undergrad Researcher · Northeastern AI Lab · Sep 2021 – Jun 2024

Medical image segmentation

Reproduced U-Net for medical segmentation on hospital CT scans; matched the paper's reported Dice on a held-out set. Wrote the full DICOM preprocessing pipeline (intensity normalization, patch sampling, patient-level splits) and ran augmentation ablations for the lab's follow-up experiments.

U-NetDICOMHospital CT

Software Engineer Intern · Beijing · Dec 2022 – Feb 2023

NetEase

AI image-style-transfer service in PyTorch + Flask with a Redis + Celery async queue. Brought inference latency from 3–5s to under 1s at 800+ DAU; engagement up 60%. Integrated Stable Diffusion into the design team's workflow — feedback cycles 3 days → 8 hours.

PyTorchStable Diffusion−85% latency
TOOLS I REACH FOR

WHAT I BUILD WITH

Mobile / Frontend

React NativeExpoReactNext.jsTypeScriptEAS release

Backend / Data

Deno edgeNodePostgreSQLSupabaseRedisKafkaZod

Web3 / Fintech

SolanaEVMUSDCCross-chain bridgingCustodial cardsKYC

AI / Agentic Systems

LLM orchestrationTool useRAGEval harnessesAgent safety + HITLOpenAI / Anthropic APIsPyTorch

Infra

AWSDockerGitHub ActionsSentryPostHog
ARC

Five Years, One Line.

  1. 2020

    Started at Northeastern. AI lab, first model training before the hype cycle.

  2. 2022

    Founded Photography Studio SaaS — first taste of shipping product end-to-end.

  3. 2023

    NetEase. Stable Diffusion in production for 800+ DAU.

  4. 2024

    Top 1% of cohort. B.S. CS at Northeastern. M.S. at NYU begins.

  5. Jul 2025

    First engineering hire at Hawala. Blank repo.

  6. Now

    Live app, $3M pre-seed, AI platform 6 teams use daily.

Build with me.

Best way to reach me is email. I read everything. Reply within 24 hours unless I'm shipping.

github.com/marlon-maDownload résumé (PDF)

New York, NY · Founding Engineer

© 2026 Marlon Ma