Résumé ↓
Soham DodiaSenior Software Engineer · Easygo

I build software that stays up when 100,000 people show up at once.

Nearly 10 years of building software, mostly in Node.js and TypeScript. I was tech lead for Stake's regulated launch in Denmark, I'm one of six engineers running moon.com, on a platform of 80+ microservices at Easygo. Open to senior and tech-lead roles.

Sheet 01 · Route & specifications2017 → now
  1. Simform2017–19Client products · AU, KR, US
  2. Gambit Sports2019–20Fantasy sports · ~10M users
  3. Clipboard Health2020–21Healthcare staffing · US
  4. Ailo2021–23Proptech · Australia
  5. Easygo2023–nowStake.com & moon.com
Spec 01 · Experience~10 yrs

building software, mostly Node.js & TypeScript

incl. internship · 5 teams
Spec 02 · Peak load100K+

concurrent users kept up

IPL surge · Stake.com
Spec 03 · Leadership7 eng

led for ~6 months, shipped on time

Denmark launch · Easygo
Spec 04 · Platform80+

microservices I ship changes into

Easygo platform
Case files

Four systems, drawn the way I'd sketch them

Each one: what was broken or missing, what I did, and what changed.

01

Holding 100,000 concurrent users during the IPL

Easygo · Stake.comIncidentKubernetesDatadog
~5 s → normalp99 latency
DWG 01scroll → 100K+users LB k8s pods scaled out cachenew layer postgres 1234 ~5 snormal p99 latency match dayrest of the season 1234
Problem

During the IPL (India's biggest cricket tournament) over 100,000 people were on Stake at once. We hadn't sized for it, and p99 latency hit around 5 seconds.

What I did
  1. 1Optimized hot paths and slow queries.
  2. 2Batched per-item database calls into bulk operations.
  3. 3Added a caching layer in front of the hottest reads.
  4. 4Scaled the Kubernetes pods to match demand.

Traced in Datadog and AWS Performance Insights. After the first rough days, smooth as butter.

02

One codebase for regulated markets

Easygo · Stake.comTech lead7 engineers~6 months
On timeDenmark launch, now reused
DWG 02scroll → Stake codebaseone, shared market layer self-exclusiongambling limitsjurisdiction rules global Denmarklaunched ✓ next marketsreusing it Danish legaladvisors 123 discovery & planning build · 7 engineers code review launched on time ✓ ≈ 6 months 4
Problem

Launch Stake in regulated Denmark, where self-exclusion and gambling limits must work before go-live, without forking a large product into a separate codebase.

What I did
  1. 1Designed a pattern that lets the existing codebase run regulated markets, instead of forking it.
  2. 2Reviewed the self-exclusion and gambling-limit work myself.
  3. 3Worked with stakeholders and external Danish legal advisors.
  4. 4Led seven engineers for ~6 months, through discovery, sprint planning, coding and code review.

Launched on time. The same pattern is now being reused for other markets.

03

Agents that turn Jira tickets into pull requests

Easygo · moon.comAI toolingClaudeCloudflare Workers
In useour PM ships small tickets with it
DWG 03scroll → productmanager Jira ticket+ tag GitHubpicks it up Claudewrites code Claudereviews it pull requestopened live preview Cloudflare Workers PM ships small tickets directly, no engineer pick-up needed 1234
Problem

Small tickets waited on an engineer to pick them up, even when the change was tiny.

What I did
  1. 1A tagged Jira ticket is picked up by GitHub.
  2. 2One Claude agent writes the change.
  3. 3A second Claude agent reviews it.
  4. 4A pull request opens with a live preview on Cloudflare Workers.

Our product manager now ships small tickets directly, which frees engineers for the bigger work.

04

Routing each payment to the gateway most likely to work

Gambit Sports~10M usersNode.jsAWS SQS
62% → 90%debit-card payment success
DWG 04scroll → card paymentincoming router card typeissuing bankgateway health Paytm Cashfreepicked for this one PhonePe 62% 90% beforeafter debit-card success 123
Problem

Too many card payments were failing on Nostra Pro, a fantasy sports app with around 10 million users.

What I did
  1. 1Score each card payment on card type, issuing bank and live gateway health.
  2. 2Route it to whichever of Paytm, Cashfree or PhonePe is most likely to succeed.
  3. 3Debit-card success went from 62% to 90%.

Also built automated payouts with fraud checks and moved transactions onto SQS.

05 · Easygo · Stake.com

Buying crypto without leaving the app

Led the MoonPay, Swapped, Mesh and Onramper integrations, and added Google Maps address checks to KYC.

~10% → ~25%of deposits now in-app
06 · Easygo · moon.com

From pre-launch to live, on a team of six

Built the admin panel support, VIP and affiliate teams use daily, the invite-only beta, and moved GraphQL from Express to Fastify.

+15%GraphQL performance
Career

Nearly ten years, five teams

Online gaming, proptech, healthcare staffing and fantasy sports, with teams and clients in Australia, the US, India and Korea.

  1. 2023 — now

    Senior Software Engineer · Easygo

    The company behind Stake.com and moon.com, running on 80+ microservices; I know many of them well enough to ship changes with confidence. Tech lead for the Denmark launch, moon.com from pre-launch, regular on-call incident lead, and author of our incident-debugging guide. −20% primary DB load with read replicas.

  2. 2021 — 2023

    Software Engineer · Ailo

    Proptech used by agencies like Ray White. Inspection-app APIs and PDF reports for 1,000+ inspections with 200+ photos each, plus full-stack tenancy workflows.

  3. 2020 — 2021

    Senior Software Developer · Clipboard Health

    US healthcare staffing. Shift verification and bulk creation, an in-house SMS and email system, Stripe. Hackathon 1st prize.

  4. 2019 — 2020

    Software Development Engineer · Gambit Sports

    Nostra Pro, later acquired by Glance. Payment routing (62% → 90%), payouts with fraud checks, affiliates and vouchers.

  5. 2017 — 2019

    Full Stack Developer · Simform Solutions

    Client products for Australia, Korea, the US and more: e-commerce, social media, travel and ride-hailing, in Node.js, React, React Native and AWS. My first open-source package.

Stack · parts list

What goes into the build

Backend-heavy, comfortable across the stack.

P-01

Backend

  • Node.js
  • TypeScript
  • GraphQL
  • Fastify
  • NestJS
  • Express
  • REST
  • Microservices
  • Go
P-02

Data & messaging

  • PostgreSQL
  • MySQL
  • MongoDB
  • Redis
  • DynamoDB
  • Elasticsearch
  • SQS
  • NATS
P-03

Infra & observability

  • AWS
  • Kubernetes
  • Docker
  • GitHub Actions
  • Datadog
  • Cloudflare Workers
P-04

Frontend

  • React
  • React Native
  • Svelte
P-05

AI tooling

  • Claude agents
  • AI code review
  • Jira → PR automation
P-06

Payments & integrations

  • Stripe
  • MoonPay
  • Onramper
  • Paytm
  • Cashfree
  • PhonePe
Appendix

Open source & output

GitHub · last year
3,149 contributions

github.com/sohamdodia ↗

Next case file

Have a system that has to hold up? Let's sketch it.

Open to senior and tech-lead roles across backend, full-stack and platform engineering.