Hi there! 👋
Final year Computer Science (Cyber Security) student in Bengaluru. I like breaking things, fixing them, and then making it harder for anyone else to break them. Mostly I build security tools, hunt for bugs, and do hands-on security research.
What I've been shipping on @pranjalm37.
I'm a final year Computer Science engineering student at Manipal Institute of Technology, Bengaluru, specializing in Cyber Security. My interests sit around secure software development, offensive and defensive security, vulnerability assessment and OSINT.
My internships have put me on both sides of the same problem. I built software at Amazon, then spent time pulling it apart at PwC and Cisco doing cloud security reviews, phishing assessments, compliance mapping, and log analysis in Splunk across simulated enterprise traffic.
What I like most is turning an attack into a tool. So far that has meant a TOCTOU vulnerability research toolkit with an AST-based static scanner, a four-algorithm API rate limiter that kills check-then-consume races, and an autonomous SOC agent running a detection-to-remediation pipeline with human approval gates.
I'm Google Cybersecurity certified and CEH v13 certified (EC-Council), with 150+ DSA problems on LeetCode and 41 rooms cleared on TryHackMe. In my spare time I build bots that track stock and deals on quick commerce apps.
PwC (PricewaterhouseCoopers) · Bengaluru
Cloud security reviews, phishing assessments, and compliance automation across enterprise environments
Amazon · Chennai
Last-mile delivery optimization for the GSF Last Mile team
Cisco · Virtual
Log analysis and network monitoring across simulated enterprise traffic
Ciphrix · Sydney
Compliance research, technical writing, and frontend polish
A TOCTOU (time-of-check-to-time-of-use) research toolkit. It ships vulnerable and patched sample pairs plus a CLI race harness that spawns concurrent attacker threads, lands symlink-swap exploits, and measures the hit rate. There's also a static scanner built on Python's ast module that flags check-then-use patterns straight from source without running anything, and a live demo you can poke at in the browser.
Four rate-limiting algorithms (fixed window, sliding window log, sliding window counter, token bucket) sitting behind one FastAPI interface, with in-memory or Redis backends you can swap out. Atomic operations via asyncio.Lock and Redis Lua scripts kill the check-then-consume race. There's a live demo that visualises traffic against each algorithm in real time, and 21 pytest tests covering the concurrency guarantees and API responses (429 + Retry-After).
An async FastAPI backend for a SOC agent that runs the whole ingest, detect, analyze, decide and respond loop across 42 REST endpoints and an 11-table PostgreSQL schema, with JWT auth, RBAC, structured logging and a Dockerized multi-service setup. A 6-rule detection engine mapped to MITRE ATT&CK feeds a RAG-grounded LLM analyst (ChromaDB + OpenAI API) that scores threats from 0 to 100, then queues remediation behind a human approval gate with rollback, guardrails and an immutable audit trail. Runs fully offline and is covered by 169 tests.
Manipal Institute of Technology
Bengaluru, Karnataka, India
7th Semester · SGPA 8.61 · CGPA 7.93 · No backlogs
Delhi Public School East
Bengaluru, Karnataka, India
ICSE 93.33% · CBSE 83%
Winner
2nd Place
4th Place
4th Place
Finalist
Finalist