SIH 2026: Which cybersecurity problem statement should we choose for a 36-hour hackathon?

Hey everyone!

My team of 6 is participating in Smart India Hackathon 2026, and we're currently stuck between a few cybersecurity problem statements.

Our biggest constraint is only 36 hours to build the prototype, so we're trying to balance technical depth, innovation, feasibility, and winning potential.

We're currently considering:

1. SIH26145 – AI-Based Detection of Cyber Threats in Unidirectional IP Traffic

Passive monitoring of one-way network traffic

Detect DDoS, port scanning, DNS tunneling/DGA, C2 beaconing, encrypted-session anomalies, exfiltration

AI/ML + behavioral analysis

Live SOC-style dashboard

No decrypting traffic or interacting with the network

1. SIH26153 – AI Based Network Attack Forecasting from Network Traffic Data

Predict future network attack progression rather than just detecting attacks

Uses network traffic features and ML/deep learning

Potentially maps predictions to MITRE ATT&CK stages

More research-oriented and technically ambitious

1. SIH26106 – AI-Powered Email Threat Detection, GeoLocation and Forensic Intelligence Platform

Phishing/spoofing/BEC detection

Email header and protocol analysis

IP/domain intelligence and geolocation

Threat correlation and forensic reporting

Potential graph-based campaign/infrastructure analysis

1. SIH26155 – AI-Driven Multi-Vendor Network Security Compliance Auditor

Analyze firewall/router/switch configurations from different vendors

Map configurations against security standards

Detect misconfigurations and generate remediation commands/reports

We're leaning toward SIH26145, but I'm wondering if we're overlooking something.

For people who have participated in SIH, hackathons, cybersecurity competitions, or similar events:

Which one would you choose?

Which has the best winning potential?

Which is realistic to execute well in 36 hours?

Which one gives the best combination of technical depth + innovation + strong demo?

Would you prioritize a technically ambitious PS or one that can be polished much better within the time limit?

I'd especially appreciate opinions from people who have actually participated in SIH or judged hackathons.

Thanks!