Security researcher and self-taught AI engineer in Bengaluru. Interested in LLM red teaming, prompt injection, agent observability and interpretability.

LinkedIn · Email · Book a 15-minute call

• Zenith, founder: AI-powered security assessments.
• Vahini Technologies, founding ml engineer: built the ML pipeline for an IMU handwriting pen. Lead inventor on 2 granted Indian patents.
• Penetration testing: 29 vulnerabilities found across web, mobile and API in authorized tests.

• TAINT: prompt-injection detection for AI agents
• ARES RED TEAM: checks coding-agent sessions against a tool policy
• REFUSAL-CLIMB: refusal as a direction in model activations
• AUTOFORGE: red teaming on a fixed budget

• Immunefi bug bounty (Orca Whirlpools)
• 47th, Gray Swan indirect prompt injection challenge
• Fixes contributed to CERT Polska Artemis, DefectDojo and CDLI

• 1st, SANS NetWars Tournament Core
• 1st, OWASP AppSec Bangalore BeSec CTF
• Top 10 at Cloud Security Alliance Bangalore, SANS Veterans Day and Amazon CTFs