SHANAHAN SURESH

I build full-stack products & AI tooling.

I'm a software engineer and a Computer Science with AI graduate from the University of Nottingham. I like turning ideas into products people enjoy using, so I build across the full stack, and lately a lot of that is agentic coding. How something feels to use matters to me as much as how it's built.

See what I build Say hi
Next.js
React
Svelte
TypeScript
Python
Django
Docker
Google Cloud Platform
Terraform
GitHub Actions
MCP
FastAPI
JavaScript
Astro
Tailwind
Bootstrap
Java
Ruby
SugarCube

ABOUT

The short version

From wanting to be an inventor to building products

Long before knowing how to print my first Hello World, I grew up wanting to be an inventor. Nowadays I make stuff that live only in code, from lean bite-sized apps to massive systems. It's an awesome feeling to create from nothing.

These days, I think of myself as a product engineer more than just a programmer. I enjoy building systems, but I care just as much about the experience around them. The best products don't just work, they should feel right.

I'm also a tad obsessive over polish. If I spot a rough edge or bug, it's hard to ignore. Once noticed, it becomes my problem.

At a glance

Focus

Full-stack products & Agentic systems

Lately

Infrastructure (Terraform/AWS) & agentic coding

Off the clock

Gym, anime, and games

HOW I THINK

Principles I build by

Make it a joy to use

People remember how software feels. Every engineering design decision should start there.

Love is in the details

Small imperfections add up. The little interactions and polish are worth the extra mile to me.

Let robots handle the boring bits

Life's too short for repetitive work, that's what LLMs are for. So we can focus crafting the fun stuff.

CURRENTLY

What I'm into lately

Right now I'm diving into deployment infrastructure and DevOps. Picked up on Terraform and use it on real projects, plus I'm teaching myself AWS in my own time. Kubernetes is next on the list.

I also spend most of my days in the agentic-coding world, trying to keep pace with how fast LLMs are changing the way software gets built. It has reshaped how I work, in the office and out of it.

Tools I reach for

Next.jsReactSvelteTypeScriptPythonDjangoDockerGoogle Cloud PlatformTerraformGitHub Actions

Currently learning

AWSKubernetes

SELECTED WORK

A few things I've built

Mix of things I shipped at work, plus stuff outside it too.

Personal

Automated artwork censoring app

A desktop app that detects and censors regions of artwork automatically, using an anime-tuned pipeline (YOLO + FastSAM). It handles batch jobs, speaks English and Japanese, and is packaged so non-developers can run it without touching a terminal.

PythonYOLOFastSAM

AI-curated tech-news digest

A daily tech-intelligence digest that reads the firehose so you don't have to. It curates the news to a team's exact stack with an LLM, runs entirely from config, and scales to zero on GCP when nobody's looking.

PythonGCPTerraformLLMs

OpenAPI → MCP server + chat UI

Point it at any API's schema and it generates a set of LLM-callable tools on its own, falling back across models if one stalls, all wired into a Next.js chat interface. It lets an AI assistant pick up a brand-new tool in seconds.

PythonMCPFastAPINext.js
Corporate work

OHSE incident-reporting module

When something goes wrong on a worksite, this is where it gets reported, investigated, and put right. It's now live across 10 business units.

The tricky part was never the forms, it was the people: managers, safety officers, and investigators each need what's effectively a different app, so I built the role-based workflows to match.

I owned it from the first round of SDLC planning through to the GCP rollout, set up the team's first Playwright test suite along the way, and added an MCP server that turns the module's OpenAPI schema into tools an AI assistant can actually call.

Next.jsDjangoGCPMCP

Automated RSVP & check-in web app

Built to welcome 75,000 guests without breaking a sweat.

It's a serverless Next.js app that runs RSVPs and check-in behind a glassmorphism UI, sends confirmations over email and WhatsApp, and hands everyone a scannable QR ticket, with SendGrid and Respond.io doing the talking.

The whole stack is defined in Terraform and ships through GitHub Actions, so scaling up for a crowd that size comes down to a config value rather than a late night.

Next.jsTerraformRespond.ioSendGrid

Enterprise HR platform

This is the flagship HR platform 20,000 employees sign into, so every slow page is a delay felt 20,000 times over. I went after the heaviest modules, reworked how queries were serialized, and added bulk processing to bring page-load times down by 63%.

While there, I wrapped PII in AES-256 transparent encryption and built custom PostgreSQL workflows for the analytics team to lean on.

SvelteDjangoPostgreSQL
Academic

HeatGen

A deep-learning project I built with a team during my uni days, later published in a peer-reviewed journal (JESTEC, 2023).

PythonDeep Learning
Read the paper

BEYOND CODE

When I'm not building

Outside of work I'm deep into anime, games, and light novels, usually half-planning a trip somewhere, and in the gym four evenings a week. Ask me about any of it and I can go a while.

GymInvestingPersonal financeAnimeGamesLight novelsTravel

Always open to interesting work & projects

Let's
build.

I'm always up for interesting work and a good conversation, whether that's professional, a passion project, or something a bit weird and fun. If anything here caught your eye, my inbox is open.

Email me