📌 About the Event

Hack[AI]Thon 2.0 2026 is a premier national-level, 24-hour Data-Centric AI Hackathon presented by KVG College of Engineering (KVGCE) Sullia’s technology club, Sphere Hive, in collaboration with VRIF VTU Belagavi.

Unlike conventional hackathons that focus on brute-forcing algorithmic hyperparameters, this “Dojo” explicitly adopts a Data-Centric AI approach. Competitors are evaluated strictly on their ability to engineer, clean, audit, and optimize datasets to achieve maximum model accuracy lifts.

---

📅 Critical Timelines & Campaign Horizon

• Round 1 (Online Filtering): June 15 — July 25, 2026 [Active and Ongoing Now]
• Finalist Pool Announcement: August 1, 2026 (Top 16 Teams Selected)
• Round 2 (Offline Grand Finals): August 8 – August 9, 2026 (24-Hour Non-Stop Sprint)
• Venue (Round 2): Visvesvaraya Technological University (VTU) Resource Innovation Foundation (VRIF), Belagavi, Karnataka, India.

---

🧠 The Structural Hacking Challenges

Round 1 (Online) | Data Detective Challenge

Teams inspect a flawed, introductory dataset of 500 Labeled and 100 Unlabeled images across 5 baseline classes: Plastic, Paper, Metal, Glass, and Organic Waste.

• Objective: Find intentionally introduced bugs (label noise, duplicate files, blurred elements, class imbalances, and distribution outliers) without altering the core baseline classifier architecture.
• Deliverable: A concise, maximum 5-slide PPT presentation or a 3-page PDF audit report showcasing evidence of data anomalies and proposed systemic fixes.

Round 2 (Offline) | The 3LC Retraining Arena

The top 16 qualifying teams advance to the VRIF VTU campus to work on an expanded scale package containing 500 Labeled and 5,000 Unlabeled images.

• Objective: Utilize the 3LC.ai data-curation framework to clean, prune, and label data strategically.
• The 4-Step Production Data Loop:
  1. Analyze Embeddings: Map and evaluate distribution anomalies across data vectors.
  2. Select High-Loss Samples: Pinpoint training pain points using 3LC metrics.
  3. Label Strategically: Manually correct bounding details or labels to break inference ambiguities.
  4. Retrain & Measure: Execute final model retraining over the optimized data subset and report precision lifts.

---

📥 Incentives, Perks & Prize Pool

• Financial Rewards: A combined cash bounty pool, sponsor awards, and tech goodies valued at ₹35,000.
• Elite Stack Access: Hands-on operational training with enterprise tools like 3LC.ai and free .xyz domain registration vouchers for all builders.
• Accommodations: On-site food, refreshments, and 24-hour workspace access are provided at the VTU campus during the offline final round.
• Target Profiles: Open to all students, software designers, machine learning developers, and AI engineers across India. Accommodates multi-disciplinary squads up to a standard team formation.

---

📧 Official Direct Coordination Channels

• Technical Support Mailbox: Connect directly with the Sphere Hive organizing desk via spherehive@kvgce.ac.in.