I suppose you're here to gaze upon my genius. Very well, try not to drool on the repositories.
The primitive tools I am forced to use to build things you couldn't even conceptualize.
- Languages of Choice: C, Python, JavaScript
- Frontend Peasantry: Next.js, React.js, TailwindCSS (yes, I tolerate CSS)
- Backend Sorcery: Node.js, FastAPI
- Data & Domination: PostgreSQL, Redis
- Infrastructure (a.k.a. Making Clouds Obey): Azure, Docker & Docker Compose, Celery, Nginx/Caddy, Linux (Ubuntu), POSIX
- Dark Arts: SSL/TLS, DNS, CORS, JWT/JWKS Auth (securing things so even you can't accidentally break them)
I have deployed a distributed, containerized, cryptographically secured remote code execution platform on a 1GB RAM Azure VM β because constraints are for people who lack imagination.
It handles untrusted user code in isolated Docker containers, queues execution jobs via Celery + Redis, streams real-time state transitions to a React terminal UI, and validates JWTs before it even thinks about running your code.
In other words: I built an entire cloud execution pipeline so you could paste your broken hello_world.py somewhere safe.
π OopsEngine β Try not to break it. I've made that surprisingly difficult.
Custom Unix Shell | C, Linux Kernel API, POSIX Systems
Yes, I built a Unix shell in bare-metal C. Not because the existing ones don't work β but because I prefer speaking to the Linux kernel directly. Middlemen are for people who don't know what execvp() does.
- Kernel Whispering: Full Unix shell via raw system calls β
fork(),execvp(),waitpid()β dictating process lifecycles across a persistent REPL becausebashdidn't ask for my permission to exist and neither did I. - Flawless Orchestration: Multi-pipe parser for arbitrarily chained commands (
ls | grep | wc -l). Dynamically allocated file descriptors, IPC across 3+ concurrent child processes, zero deadlocks. Obviously. - Stream Hijacking: Standard stream redirection via
dup2()andopen()β both>(overwrite) and>>(append) β because I route process outputs exactly where I want them. - Custom Tokenization: Two-phase tokenizer engineered from scratch to handle nested pipe contexts where
strtokcollapsed under the pressure of its own design flaws. You're welcome, C standard library. - Memory Mastery: Zero memory leaks. Zero segfaults. 500+ REPL test iterations. Every heap allocation tracked and freed manually. The Valgrind report is cleaner than your commit history.
- LeetCode: 180+ problems solved in Python, including the full NeetCode 150 β arrays, trees, graphs, DP, sliding window. A cognitive stretching exercise I squeeze in between building things that actually run in production.
- Open Source: Both projects are publicly maintained on GitHub with documented READMEs. Yes, I write documentation. I know. Shocking.
Bachelor of Computer Applications (BCA) β Dehradun Institute of Technology (Expected May 2027)
Coursework includes: Data Structures & Algorithms, Operating Systems, DBMS, Computer Networks.
Note: The degree is technically in progress. The competence is not.
If you absolutely must interrupt my work, you may attempt to reach me here. I make no promises regarding response latency.
- π Portfolio
- πΌ LinkedIn
- π¦ Twitter/X
- π§ pksingh69313@gmail.com (reserved for opportunities sufficiently worthy of my attention)
