P.04 — AI · Scale · Live system
BUILD2GETHER AI–SDG Hackathon
Lead engineer for the platform behind an international AI hackathon series — about 5,000 students across three live events.
- Role
- Lead engineer · Platform, Azure deployment, live delivery
- Year
- 2025 — 26
- Company
- Build2Gether · Freelance engagement
01 — Overview
Overview
BUILD2GETHER was built in partnership with Build2Gether and researchers affiliated with ETH Zürich and NTU Singapore. Teams proposed projects supporting the UN Sustainable Development Goals and received AI feedback on their proposals.
As part of the research, half of the teams got feedback on improving their ideas to fit the predefined goals; the other half got feedback identifying redundancies with projects already under way. The second group came up with a wider range of more innovative ideas.
I owned platform development, Microsoft Azure deployment and live operational delivery.
02 — The problem
A 24-hour live event with close to 2,000 participants in nearly 500 teams leaves no room for downtime.
The platform had to hold up under peak load, deliver the right kind of AI feedback to the right teams, and be operated live — three times.
03 — The system
Running a live platform at event scale
Every event followed the same discipline: prove it under load, deploy it, then operate it live for the whole event.
- 01
Load test
Locust load testing before every event.
- 02
Deploy
Microsoft Azure deployment and configuration.
- 03
Teams
Participants organised into teams — ~500 at the largest event.
- 04
Proposals
Teams submit project proposals for the UN SDGs.
- 05
AI feedback
Two feedback conditions, as designed by the research team.
- 06
Operate live
Live operational delivery through the event.
04 — My contribution
Team, and what was mine.
Team
- Build2Gether and researchers affiliated with ETH Zürich and NTU Singapore
- Freelance engagement, working directly with Build2Gether
My contribution
- Lead engineer for platform development
- Microsoft Azure deployment
- Locust load testing of every platform before each event
- Live operational delivery across all three events
05 — Technical depth
What it's built with, layer by layer.
- Cloud
- Microsoft Azure
- Testing
- Locust load & performance testing
- Delivery
- GitHub Actions · Azure DevOps · Multi-environment deployment
- AI
- AI feedback on proposals
06 — Gallery
In pictures.
07 — Outcome
Where it landed.
~5,000
Participants
across three live events
3
Live events
run on the platform
~2,000
Largest event
participants at Sairam Institutions
~500
Teams
at the largest event
Three live events, about 5,000 students, and the largest — close to 2,000 participants in nearly 500 teams — ran live after Locust load testing.
The hackathon is described by name in Harvard Business Review (July 2026).
Featured in · Harvard Business Review · July 2026
“Design AI Systems That Actually Strengthen Human Reasoning”
The article describes the BUILD2GETHER AI–SDG Hackathon by name — the platform behind it is the one I led engineering and live delivery for.