Hi, I’m
Aaryaman
⚠️ Warning: User may be mood-oriented. Sarcasm included by default.
A developer who likes building things
that feel alive.
Writes code that reacts. Cares about how things feel, not just if
they work. Obsessed with details most users won’t notice — but will feel.
Most of what I build starts as curiosity and ends as a project. I tend
to obsess over details, break things on purpose, and rebuild them
better the second time.
Detail-oriented.
Mildly over-engineering things on purpose.
Still learning. Always shipping.
¯\_(ツ)_/¯
Languages
- JavaScript | Python | C | SQL
Frontend
- React.js | Next.js | HTML | CSS | Tailwind CSS
AI / Data / Systems
- Machine Learning (basics, applied)
- MLOps – pipelines, deployment workflows
- API integration (OpenAI, external services)
- Data processing & visualization
Creative / Interactive
- Three.js / WebGL — particles, 3D, chaos
- Creative coding — visuals that respond
- Animation — subtle > flashy
- Figma — prototyping interactions, not just screens
Tools & Workflow
- Git & GitHub — versioned chaos
- VS Code — heavily customized
- Debugging — break → observe → fix
Selected Modules
A GPU-Accelerated Bioinformatics Engine with a high-performance pipeline using CUDA for protein data processing. High-throughput Proteomics Engine executing 3B-parameter ESM-2 models with custom Triton kernels. Features a dual-branch MLOps pipeline achieving ~34,500 proteins/hr with 100% data integrity via FP32 overflow guards.
Automated MLOps infrastructure for real-time NIDS (Network Intrusion Detection). Engineered a modular 5-stage pipeline orchestrated by Apache Airflow and MLflow to automate the full lifecycle—from ingestion and transformation to model registration. The system implements automated Quality Gates to benchmark Random Forest models before promotion, ensuring high-reliability deployment via FastAPI and Docker for zero-latency threat detection.
A smart subscription tracker that helps you understand where your money goes — monthly, yearly, and over a lifetime. No forgotten trials. No surprise renewals. Just clarity.
Interactive hand-gesture controlled particle void using Three.js & MediaPipe.
A minimal Chrome extension that introduces a pause between impulse and intent, helping users rethink actions before they commit.
An experimental web app that captures and visualizes the internet’s current mood using live data and sentiment signals.
RAG-driven Health AI for clinical risk screening and triage. Built with Next.js and OpenRouter, it features an ML-powered emergency gateway to prioritize critical cases and contextual memory for protocol-grounded patient assessments.
Open to internships, hackathons, and weird ideas.
I reply fast unless the Wi-Fi gives up or my mom calls.
Miscellaneous
Things that didn’t fit anywhere else. So I put them here.