Ayush Basral

I build software.

Backend systems, AI products, weird side projects, and occasionally companies.

GoDaddy · ex-Amazon · Pictory · Third Unicorn

Open to one serious build

Selected work

01 — 2024 — now

GoDaddy

Identity infrastructure at scale.

Software Engineer II

godaddy · event fan-out

Identity and profile services sit under everything else, so their failure modes are everyone's failure modes. The hard part is never the happy path — it's what happens when a provider is down, a message arrives twice, or a certificate expires quietly at 3am.

  • Identity and profile platform services
  • Event pipelines on Kinesis, with replay and dead-letter handling
  • Compliance and sanctions screening, off the request path
  • Certificate lifecycle automation — renew, adopt, retire, idempotently
  • Internal AI tooling: a graph memory layer that gives Claude and Cursor organisational context, in production for my team
  • Java
  • Python
  • AWS
  • Kinesis
  • ECS
  • Lambda
  • PostgreSQL

02 — 2023 — 2024

Pictory

AI that made it past the demo.

Software Engineer

hivemind · context flow

Built the LLM pipelines behind automatically generated video — scripting, captioning, summarisation, metadata — and the public APIs other teams shipped on top of them. The work was less about the model than about everything around it surviving real traffic.

+25%
User engagement
  • Node.js
  • Python
  • OpenAI
  • AWS

03 — 2023

Third Unicorn

10,000 people. One cricket match. Don't go down.

Senior Software Engineer

cricpe · request path

Scores arrive in bursts, every leaderboard recalculates at once, and everyone refreshes at the same moment. Leaderboard computation and PDF generation were happening inside the request cycle, so the slowest work ran at the busiest time. Moving both onto Redis and RabbitMQ took them off the request path entirely.

10k+
Concurrent requests
92%
Faster leaderboards
60%
Server performance
  • Node.js
  • Redis
  • RabbitMQ
  • MongoDB
See it live→

04 — 2021 — 2022

Amazon

Recommendations at marketplace scale.

Software Development Engineer I

Distributed Java services behind Amazon Fashion, at millions of requests a day. A recommendation system for discounted offers on product detail pages, and a rebuilt shade selector that put every colour side by side instead of behind a dropdown.

+20%
Sales, discounted offers
+20%
Detail-page engagement
  • Java
  • AWS
  • Recommendations
  • A/B testing
See it live→

Experience

  • 2024 — nowGoDaddySoftware Engineer II
  • 2023 — 2024PictorySoftware Engineer
  • 2023Third UnicornSenior Software Engineer
  • 2021 — 2022AmazonSDE I
  • 2018 — 2021FreelanceSoftware Engineer
See full résumé→

Things I've built

Founder · beta

Eigon

eigon · run log
  1. 01 intent received✓
  2. 02 plan generated✓
  3. 03 risk: medium!
  4. 04 approval required⏸
  5. 05 change executed✓
  6. 06 deployment verified✓
  7. 07 coffee still warm☕

An agentic infrastructure platform. It turns intent into a structured plan, risk-classifies every action, stops for human approval before anything mutating, executes through least-privilege tools, and keeps durable state so a failed deploy reverses instead of stranding half a stack. Go control plane, BYOC on AWS.

  • Go
  • AWS
  • Terraform
  • PostgreSQL

Open source

echo-mem

Long-horizon memory for agents. Facts are edges between entities rather than flat vector rows, old memory consolidates into summaries instead of piling up, and every change is auditable. Postgres, pgvector and Apache AGE underneath, exposed over MCP — extraction runs in the calling agent, so a write costs no LLM calls. It's also my M.Tech research, shipped.

  • Python
  • PostgreSQL
  • pgvector
  • Apache AGE
  • MCP

Technical lead

Dugout

dugout · traffic shape

A sports platform, taken end to end from React Native client through Node and Postgres to the infrastructure and the incidents. WebSocket chat, feed ranking with diversification, notifications, offline sync — all built for traffic that does nothing for an hour and then arrives at once.

  • React Native
  • Node.js
  • PostgreSQL
  • Redis
  • WebSockets

Community · Chandigarh

IXC Commons

A hands-on community for people building with AI, which I help run. Small rooms, open laptops, everyone brings a real problem and leaves with something that runs.

Founder

Ateleir Tech

A small engineering agency for early-stage teams that need something built and running, rather than specified.

The archive — 12 older things

About

I'm Ayush. I started writing software seriously around 2019 and somehow haven't stopped.

I've built distributed services at Amazon, video and LLM infrastructure at Pictory, survived match-day traffic at Third Unicorn, and now work on identity systems and internal AI tooling at GoDaddy.

Outside work I have a bad habit of starting things — products, experiments, communities, occasionally a company. Some of them even become real.

I'm especially interested in what happens when AI systems stop being demos and have to survive real users, real infrastructure and real failure modes. That's most of what I think about right now, and it's what my M.Tech research is on.

I work mostly with Go, TypeScript, Python, Java, PostgreSQL, Redis and AWS — and whatever else gets the thing shipped.

/misc

Coffee
Yes
Timezone
IST — and I'd rather not meet before 10
Favourite bug
The one someone else introduced
Currently exploring
What agents remember, and what they should forget
Commits after midnight
More than is medically advisable
Rabbit hole
Something I will probably turn into a product

Got something
interesting?

ayushbasral2000@gmail.com

+91 98163 79471 — 90% chance I don't pick up.

© 2026 Ayush Basral