My degree is Civil Engineering, not CS. That's mattered less than people assume.
I taught myself the rest, from first principles. Structures had already taught me the part that matters: where things break, what fails first, what a system looks like under load.
Now I build across ML research and backend engineering, happiest at the boundary between them, where the maths has to be right and the system has to actually run. I like shipping real things: a package on PyPI, a model that trains, a backend that stays up.
Maths is the thread through all of it.
A few things I've built.
Spanning machine-learning research, backend infrastructure, and low-level systems.
click a node to open the project
A trajectory of gradient descent, with momentum, overshoot, and convergence.
Scroll, and θ rolls down the loss surface.
Research-grade ML pipelines, backend and AI infrastructure for deep-learning training and evaluation.
Amazon ML Summer School 2026, top 2.2% of 140k nationwide. 1st, Silicon Chip Smackdown AI Poker · 3rd, AI Chatbot Hackathon.
CGPA ~8.0. Civil Engineering with a Minor in IT, the CS core: DSA, operating systems, networks, databases.
The tools I actually reach for.
Grouped by where they live in the stack. Hover a row to pause it.
Code that lives in the open.
A keyboard-driven terminal music player in Python, published to PyPI. A reactive Textual TUI over a C-bound Miniaudio backend, with O(1) library indexing for near-instant boot.
∇ context-kernelACK: memory middleware for terminal AI agents. Collapses noisy logs and tracebacks to their signal (~95% token compression, 100% fidelity) and archives the full stream to a searchable store.
Σ gc-suaeA geologically-constrained autoencoder for unsupervised lunar mineral mapping from Chandrayaan-2 IIRS, fusing hyperspectral, terrain and iron maps through spatial cross-attention.
λ getstreamA terminal-first API debugging client, a lightweight Postman alternative with a React dashboard. Bypasses browser CORS with raw TCP from a zero-dependency Node backend.
∞ more on githubDocumented projects, experiments, and the occasional rabbit hole: ship-route RL, deepfake detection, crowd counting, a C TCP server, and more.
Notes from the boundary of research and engineering.
Let's build something that converges.
Open to software engineering and ML/AI internships worldwide, and to good conversations about systems, research, and everything in between.