Building systems that work before making them clever.

I'm Spurti — a Computer Science undergraduate building software at the intersection of backend engineering, distributed systems, and intelligent applications.

I like understanding how things work beneath the abstraction — whether that means tracing a request through a distributed service, understanding what a model actually learned, or discovering that the obvious solution isn't always the right one.

KLE Technological University · Computer Science · 2027

At a glance

CVPR 2026 Workshop

NTIRE 2026 · Published in the Workshop Proceedings

Amazon ML Summer School 2026

Selected among ~3,000 from 134,000+ applicants

Backend Engineering

CDPI · DIGIT 3.0 · Go · Kafka · PostgreSQL

Selected Work

Three projects I keep coming back to.

01

NTIRE 2026

Building an image enhancement model under a strict computational budget.

The challenge was not simply to make an image enhancement model perform better. It was to make it small enough to be useful. Working under a 1 MB model constraint forced me to think much more carefully about architecture, parameter efficiency, and which improvements were actually worth their computational cost.

~250K parameters · <1 MB model · Rank 14 / 1,500+ · CVPR 2026 Workshop

PyTorchComputer VisionEfficient MLONNX
02

E-Sannidhi

An offline-first telemedicine platform designed for low-connectivity environments.

Built during the Smart India Internal Hackathon 2025, E-Sannidhi was designed to connect healthcare workers and patients in places where reliable connectivity couldn't be assumed. The system had to let users continue working offline and reconcile data when connectivity returned, making synchronization and conflict handling central to the design rather than something added after the UI was built.

Rank 31 / 200+ teams · Offline-first data capture

ReactNode.jsPostgreSQLIndexedDB
03

Emotion-Aware Assistant

Exploring how emotion recognition and language models can make conversational systems more context-aware.

The system combined emotion classification with a language model, using the detected emotional state as additional context rather than treating conversation generation as an isolated language-model problem.

Transformer-based emotion modelling

PythonPyTorchNLPTransformersLLMs
Explore all projects →

Experience

I've spent the last couple of years moving between production software, backend systems, and computer vision research.

CDPIBackend Engineering2026–Present
CEVIResearch & Computer Vision2025–Present
KnitspaceSoftware Engineering2025
View full experience →

Field Notes

Short reflections on things I've learned while building.

Read all field notes →

Let's talk.

I'm always interested in good engineering problems, interesting research questions, and conversations that make me rethink something.