Software · Applied AI · Adelaide

Agrim Sharma

Final-year Computer Science student at Adelaide University, majoring in Artificial Intelligence. I build software that solves problems, put AI to work inside them, and ship it so real people actually use it.

Interning on voice AI at Aurivox · open to graduate roles from December 2026

agrimsh22@gmail.com

Projects

Problem, approach, and impact. The stack comes second.

Pacific Village Explorer

Top 12 finalist · 2026 Humanitarian Innovation Hackathon

Problem

A village council deciding where to rebuild after a flood, where to move a school, or how to farm ground turning saline has no way to see its own coastline in twenty years. The digital twin platforms that answer that cost over $10,000 per site and need technical staff villages do not have.

Approach

Built in 44 hours at the University of Sydney with a team put together on day one, across three universities. A coordinator picks their village, drags a timeline from 2026 to 2075, and watches houses, wells, farms and sacred sites change status as the water rises, with adaptation options for every asset at risk.

I led the AI recommendation engine. Each prompt carries the village's real context into the Claude API: its population, budget, cultural constraints, sacred sites and available labour, alongside IPCC AR6 and SPREP climate data. Guardrails keep every recommendation tied to what that council can actually do, instead of generic climate advice.

Impact

Top 12 of 75 submissions from 110 teams, and the only Adelaide University student to reach the Grand Final, where we pitched in Sydney on 19 August. Judged by a panel including Sir Peter Cosgrove, Canva, Commonwealth Bank's AI Scientist and the Pacific Region Infrastructure Facility. Under $500 per village against $10,000+ for the alternatives, on any phone from the last five years.

Read the case study

Source Finalist announcement

React · Vite · Tailwind · Leaflet · Claude API · Vercel

Agrim Sharma with his two teammates at the Humanitarian Innovation Hackathon — The team

Whiteboarding the app architecture during the hackathon — Whiteboarding the build

The three-university team building during the 44-hour hackathon — 44 hours in

MetaPlay

Live · Built 2025, deployed 2026

Problem

Gamers track what they play across scattered notes, spreadsheets, and memory. Our brief was a single place to discover games, build a collection, and review them, backed by real game data rather than a static seed list.

Approach

I worked in a five-person team and owned the front-end and back-end integration. I built the authentication layer, with bcrypt-hashed email and password login plus Google sign-in through Passport, across separate admin, user and guest permissions, and designed the nine-table relational schema behind collections, reviews, friends and groups.

When the course ended, so did everyone else's involvement. A year later I came back to it on my own, traced the broken OAuth and deployment configuration, wired it to live RAWG game data, and put it into production.

Impact

It runs live today with personal collections (wishlist, currently playing, completed), reviews, friends and groups, and an admin dashboard, pulling real-time data from the RAWG API. The deployed version of a team project that I took the rest of the way and shipped myself.

Read the case study

Live demo Source

Node.js · Express · Vue.js · PostgreSQL · Passport.js · RAWG API

Freelance · 2026

Adelaide Rising Stars

A mobile-first website and enquiry system for a tennis academy coaching 100+ players. My first paid client build.

I scoped it with the owners, quoted it, and talked them out of self-service booking, which would have overfilled sessions. Delivered in three weeks.

Nuxt · Tailwind · Nuxt Content · Netlify Forms

Read the case study

Client project · live link coming soon

Next.js · 2026

Portfolio AI Assistant

The site you are on, and the assistant in the corner that only answers from a knowledge base I wrote.

Streams answers from the Claude API through my own server route, so the key never reaches the browser, with rate limits and injection refusal on the endpoint.

Next.js · TypeScript · Tailwind · Claude API · Vercel

Read the case study

Source

Java · 2026

Concurrent Speech-to-Text Service

A Spring Boot web service that transcribes speech through OpenAI's API and passed all eleven automated functional tests.

Built for many requests at once: thread-safe request handling and usage stats, plus admin and stats REST endpoints implemented against an OpenAPI spec.

Java · Spring Boot · OpenAI API · REST · OpenAPI

Read the case study

Private repo (coursework) · walkthrough on request

Java · 2026

GPS Tracker Dashboard

A real-time dashboard that merges ten live GPS streams without a single piece of shared mutable state.

Built with functional reactive programming, so location updates flow through composable streams instead of event handlers mutating state. Covered by 19 automated checks.

Java · Sodium FRP · Swing

Read the case study

Source

Python · 2026

Pathfinder

BFS, uniform-cost search and A* from scratch on terrain where climbing costs more. A* expanded 41 nodes where BFS needed 112.

Three classic search algorithms on elevation-aware grids, compared on the same maps to show exactly what a good heuristic buys you.

Python · BFS · Dijkstra · A*

Read the case study

Source

C++ · 2024

Virtual Restaurant Simulator

A restaurant simulation designed around clean class hierarchies, so new staff roles extend it without rewrites.

Fourteen classes and 1,200+ lines of C++, with inheritance and virtual methods letting each staff role define its own behaviour through a shared interface.

C++ · OOP · Makefile

Read the case study

Source

Leadership

Adelaide University Tennis Club · President, Jul 2024 to Mar 2026

A dormant club to Club of the Year in eighteen months.

I took over a club that had gone quiet and rebuilt it from scratch. I led the merger of two university tennis clubs during the Adelaide and UniSA consolidation, co-authored the new constitution, brought together an eight-member committee, and won grants for facility upgrades. The real challenge was getting two groups who didn't know each other to trust the process and work as one.

Adelaide University Sport Club of the Year cheque presentation — Club of the Year · Adelaide University Sport, 2025

2 → 1
clubs merged: $7,000+
grants secured: 10 → 100+
active members: Winner
Club of the Year 2025

The challengeRunning our side of a two-university club mergerRead it

Setup

Adelaide and UniSA were merging, so every affiliated club had to merge too. I was president of the club that was ahead on membership and results, so I ran our side of it. That meant recruiting and bringing the right people across, aligning two committees that each ran their own social and competitive tennis, leading the merged constitution, and keeping every stakeholder comfortable, the University most of all, with as few conflicts as possible.

What I did

I started by working through what each committee actually wanted and where the two overlapped, then led the merged constitution until both sides were happy to sign it. The hardest piece was the coaches. They were two independent businesses, each loyal to a different club, and any shared model had to close the gaps where money could quietly leak out of the club. We talked through several models over a lot of meetings and settled on a three-month trial with both coaches running sessions together, so the real problems would surface early and we could fix them before locking anything in. Alongside that I ran member feedback to keep people bought into the change, and took on the practical handover myself: the facilities, the nets and fencing, the spare rackets, and the full asset register.

Result

One merged club running smoothly, both coaches retained on a model that actually worked, the members kept happy through the transition, and Club of the Year in the same year.

Experience

Where I've led and worked

Across leading a committee, coaching more than fifty clients of every age and background, and working a busy retail floor, the common thread is working with people. It's made me good at reading and communicating with people from all walks of life, and comfortable fitting into any environment, including the Australian workplace I want to build my career in.

Aug 2026 to Nov 2026

Software Engineering Intern, Voice AI, Aurivox · Viemo Capital & Consulting

Chosen as one of four interns from eleven shortlisted students for a 50-day placement with a startup building a voice AI product. I work across the speech-to-text, language model and text-to-speech stack of a voice-first meeting assistant, reporting to the founder and taking an open brief through to a working prototype. It is the first time I have shipped AI features against a founder's brief rather than a marking rubric.

Jul 2024 to Mar 2026

President, Adelaide University Tennis Club

Led the club revival covered above: merger, constitution, committee, grants, and the award.

Mar 2024 to Present

Assistant Head Coach, Adelaide Rising Stars Tennis Academy

I coach 10+ sessions a week at Tea Tree Gully Tennis Club for more than fifty clients, from juniors to adults, one-on-one and in groups. I run junior and performance squads of up to thirty players in a two-hour block, setting the drill plan and directing assistant coaches across four courts. It has made me good at explaining the same idea a few different ways until it clicks, and at keeping a big group moving on one plan.

Feb 2024 to Oct 2024

Retail Assistant, IGA Supermarkets

Part-time customer service across two stores while studying full-time. Busy retail floor, steady standards, and a lot of practice juggling work and study at the same time.

Mentorship

Career Access Mentoring Program · Adelaide University

Credential

A structured 14-hour mentorship through Adelaide University, spread across meetings over four to six months, mentored one-on-one by an APT Analyst at Google Threat Intelligence. We have stayed in touch for more than a year since. He ran me through a full mock Google interview loop, from the online assessment to the technical, behavioural and HR rounds, and was direct about exactly where I was weak and what to work on. He showed me how each round actually works, how to structure an answer, and which of my own experiences to draw on for it. We also covered how engineers at that level approach software craft and problem-solving, and where my strengths fit best. We met in person for the first time at Google's Sydney office, where he gave me a tour, and he still reviews my thinking whenever I am making a decision about my career.

University of Adelaide Certificate of Completion for the Career Access Mentoring Program

Visiting my mentor at Google Sydney

Agrim Sharma beside the neon Google sign in the Google Sydney office — Google Sydney office

Certifications

Azure Fundamentals (AZ-900) · MicrosoftCredential2026
Technology Job Simulation · Deloitte AustraliaCredential2026
Data Analytics Job Simulation · Deloitte AustraliaCredential2026
Partnering with AI in the Workplace · DatacomCredential2026

About

I build the way I play tennis, which is to say I care about the boring part. Fifteen years on court teaches you that the good shots come out of a lot of unglamorous, repeatable work, and that following through when it stops being fun is most of the game. I still coach, and watching a player finally land a shot on the fortieth try feels a lot like watching a stubborn build go green.

Leading the tennis club turnaround is where that became how I think about software. Getting a group of people who didn't agree at first onto one plan is not far off getting a tangled codebase into a state the next person can actually pick up. Both come down to follow-through, and to still caring once the interesting part is over.

When a humanitarian AI hackathon came up in Sydney, I backed myself, booked the trip interstate on my own, and built for 44 hours with people I met that weekend. We finished in the top 12. Unfamiliar rooms on short deadlines turn out to be where I do some of my best work.

Agrim Sharma playing a forehand — Match play · Adelaide

A few things I hold to, that my projects show

I would rather ship something real people use than perfect a prototype nobody sees.

MetaPlay began as a group assignment. I took it the rest of the way on my own and put it in production, so today it is a live product instead of a repo I describe in interviews.

I build so the next person can extend it without a rewrite.

The Restaurant Simulator was designed around exactly that: adding a new kind of staff means adding a class, not editing the code that already works.

I reach for the simplest thing that solves the real problem.

Not the most impressive one. The point is the person on the other end, not the framework I get to use.

Right now I am in my final semester, interning on a voice AI prototype at Aurivox, and finishing the website for the tennis academy I coach at. I graduate in December.

AI major
Computer Science: Dec 2026
Graduating: 15 years
On court: Adelaide
Based in Australia

Contact

I'm looking for graduate roles in software engineering, AI, data, and analytics. Email is the best way to reach me. I'm based in Adelaide and happy to relocate within Australia.

Email me

Currently in Australia on a student visa with work rights for internship and part-time roles. Full, unrestricted working rights from 15 December 2026 (graduation), eligible for the subclass 485 Temporary Graduate visa. Available for graduate roles from December 2026.