Spurti

Experience

What I built, what I learned, and how working on real systems changed the way I approach engineering.

01

Student Developer · Backend Engineering

Centre for Digital Public Infrastructure (CDPI)

eGovernments Foundation & Infosys · Mar 2026 – Present

Go·Kafka·PostgreSQL·Docker·Kubernetes·Distributed Systems

At CDPI, I work as part of a 4-member backend team on the migration of DIGIT 3.0's municipal services from Spring Boot to Go. The challenge is not simply rewriting services in another language — it is understanding an existing distributed system well enough to change its implementation without changing the behaviour that other services depend on.

I spearheaded the migration of two Spring Boot microservices to idiomatic Go, taking major ownership of the implementation. This work reduced container memory from ~600 MB to <100 MB per pod and cold-start time from 15 s to <100 ms. I also designed goroutine-based Kafka consumers using bounded channel semaphores, increasing event throughput by >200% over the Java thread-pool baseline.

On the data layer, I replaced JPA/Hibernate ORM with optimized sqlx executions and custom struct mappers, achieving an ~50% reduction in query latency. I also worked with PostgreSQL schema design and targeted indexing for workflow state management, alongside Docker-based containerization and Kubernetes deployment.

What I worked on

  • Service migration — Spring Boot → Go · 2 microservices
  • Messaging — Kafka · goroutines · bounded concurrency · >200% throughput
  • Data layer — PostgreSQL · sqlx · custom mappers · targeted indexing
  • Infrastructure — Docker · Kubernetes · distributed service deployment

Engineering impact

Container memory~600 MB → <100 MB per pod
Cold start15 s → <100 ms
Query latency~50% reduction
Event throughput>200% over Java thread-pool baseline
02

Project Intern · Kalpith Project

Centre for Visual Information Technology (CEVI)

Kalpith Project · Oct 2025 – Present

PyTorch·Stable Diffusion·NeRF·Point Clouds·Computer Vision·Generative AI

My work at CEVI spans two related problems: generating usable 3D assets from text-driven imagery and reconstructing missing structure in degraded heritage data. Working with incomplete geometry has made me think beyond whether a model produces a plausible output — a reconstruction can look convincing numerically while still being structurally wrong.

Text-to-3D Generative Pipeline

I am building a multi-stage pipeline that moves from text prompt → Stable Diffusion image synthesis → NeRF-based 3D reconstruction → mesh generation → motion synthesis, with the goal of producing animatable 3D assets for heritage and architectural content.

Pipeline

Text Prompt
→
Stable Diffusion
→
Generated Image
→
NeRF
→
Mesh

Heritage Monument Reconstruction

In parallel, I have worked on restoring structural details in degraded heritage monuments from incomplete LiDAR scans. I worked with transformer-based point-cloud completion models including PoinTr and AdaPoinTr, along with preprocessing pipelines for sparse point-cloud inputs.

Research status: Manuscript in preparation

Evaluation

Chamfer Distance0.0142
F-score78.4%
Normal Consistency0.74
03

Software Development Intern

Knitspace

Jul 2025 – Sep 2025
Python·Web Development·Search Systems·Data Processing

At Knitspace, I worked on an internship project focused on building a web-based search experience over 21,000+ product records. The project combined product-data search and processing with a themed web interface designed around the search experience.

I worked on the search and processing side of the application while also shaping how the search experience was presented through the website. The project used an Animal Farm-inspired theme, turning the search interface into a more experimental and engaging way of interacting with the underlying product data.

What I worked on

  • Search application — Search over 21,000+ records
  • Data processing — Preparing product data
  • Web interface — Building the search UI

Leadership & Other Experience

Research & Documentation Lead

Word's Worth Club · 2023 – Present

As Research & Documentation Lead, I have coordinated research and documentation across 14 workshops and 8 competitions, while also representing KLE Technological University at 2 intercollegiate events.

Projects·Field Notes·Achievements