Opportunity brief
About the Opportunity:
AI Gladiator is a four-round, bracket-style team competition that tests participants across the full spectrum of applied AI/DS skills — general knowledge, generative image prompting, AI-assisted coding, and startup ideation and pitching. Teams face off head-to-head and are progressively eliminated across rounds, culminating in a live pitch judged by a panel.
Guidelines:
- Open to all students eligible under the club's participation policy; teams must be finalised at the time of registration.
- Team participation only — no individual entries. Team size: minimum 2, maximum 4 members.
- Inter-college team members are allowed.
- Inter-specialisation team members are allowed.
- Round 1 draws from a question bank of 20-30 mixed-difficulty questions.
- Round 2 and Round 3 each use one target (image / output) per team match-up.
- Round 4 assigns each team a real, existing industry problem statement.
- Teams must report on time for each round; late arrival may result in forfeiture of that round, at the organisers' discretion.
- Any attempt at unfair means (sharing answers across teams, using disallowed tools, etc.) will lead to disqualification.
- The organising committee's decisions on scoring and eliminations are final.
Round-by-Round Rules:
Round 1 — SpeedGPT
- Opening round to test teams' general AI/DS knowledge and warm up the bracket.
- Each team faces its bracket opponent in a head-to-head quiz.
- Rapid-fire format: short questions with a tight time limit per question (10–15 seconds).
- Buzzer-based or simultaneous-answer format, using a buzzer system or a platform such as Kahoot/Quizizz for scale.
- Question bank: 20-30 questions of mixed difficulty.
- Round duration: 10–15 minutes per match-up.
Round 2 — Pixel Prompting
- Teams are shown a target image on the projector.
- They must write one prompt (or a maximum of 2 attempts — to be fixed and announced upfront) to recreate the image using an AI image generator.
- Output is compared side-by-side with the target image on screen.
- Scoring: visual similarity (judged), with bonus points awarded for fewer words used in the prompt.
Round 3 — Coding Circuit
- Teams are shown only a target output (console output, a specific print statement, a computed result, a small game state, etc.) — never the underlying code.
- Teams prompt an AI tool (Assigned AI Tool) to generate Python code that reproduces that exact output.
- Code is pasted into the team's own environment (Replit, Jupyter, VS Code, or whichever is fastest for them) and run live.
- The first N teams to produce the exact matching output win, ranked by timestamp of submission.
Round 4 — Brains to Bucks
- Each team is assigned a real, existing industry problem (e.g. logistics delays, farmer produce wastage, healthcare access in tier-2 cities) — not an invented case.
- Teams get dedicated prep time to research and ideate a startup solution.
- Each team pitches in a fixed time slot: 5–7 minute pitch followed by 2–3 minutes of Q&A.
- Judged by a panel; where possible, an external founder or mentor is invited to raise the stakes and credibility of judging.