Systems / Product / Research

Kunwarbir SinghPadda

Systems & Product Engineer

I build Neeh, an open-source digital-ink SDK, and work across production products, LiDAR workflows, and applied research.

  • Digital ink
  • Product systems
  • Geospatial data
  • Applied research

About.

I'm a systems and product engineer who likes making difficult technical work understandable, testable, and useful.

My work spans the open-source Neeh SDK, production product engineering, LiDAR and point-cloud workflows, computer vision, and empirical software research. I'm most useful where a problem crosses boundaries: from native code to Python, mobile UI to backend state, or research questions to reproducible analysis.

SDK
Developer tools with stable interfaces, validation, and reproducible evidence
Systems
Mobile, backend, data, and operational paths considered as one product
Spatial
LiDAR, point-cloud quality control, and Python geospatial workflows

Tools I use

PythonC++CTypeScriptReact NativeNode.jsPostgreSQLRedisDockerCMakeGitHub ActionsLiDARPoint CloudsPyTorch

How I work

Evidence before claims

Benchmarks, tests, traces, and failure cases are part of the explanation, not something added after the implementation.

Interfaces that survive real use

I care about stable contracts, explicit state, bounded work, recovery paths, and the details that make systems operable.

Research into product

I like moving between investigation and implementation, then folding the proven seam back into a tool people can actually use.

Selected Work.

A focused selection of systems work, product engineering, spatial data, computer vision, and empirical research. Public links point to the work and evidence behind each claim.

01

Open-source systems work

Neeh SDK

A structured digital-ink SDK that turns pen strokes into stable, queryable context for assistants, analysis tools, and interoperable ink workflows.

PythonC++17C ABICMakeDigital InkCI
  • —Designed stable stroke identity, temporal and geometric analysis, and bounded context retrieval.
  • —Ships validated tool surfaces, UIM support, Python APIs, and a C++17 core with a C ABI.
  • —Backed by benchmarks, a technical paper, CI, and a public v0.2.0 prerelease.
View Neeh on GitHub
02

Backend systems reference

Subscription Platform Core

A clean-room TypeScript reference for durable subscription lifecycle processing and the failure boundaries that are difficult to retrofit later.

TypeScriptNestJSPostgreSQLWebhooksDockerCI
  • —Claims webhook events durably before any domain effect and exits duplicate delivery early.
  • —Commits subscription state, audit history, outbox work, and inbox completion atomically in PostgreSQL.
  • —Documents RBAC, HMAC verification, failure modes, limitations, and real-database integration tests.
View the systems reference
03

Geospatial data engineering

LiDAR Systems Lab

A small, auditable Python reference for LAS/LAZ metadata inspection and deterministic point-cloud quality control.

PythonLAS/LAZPoint CloudsGeospatial DataQuality Control
  • —Checks CRS, bounds, finite coordinates, scale and offset safety, classifications, and return consistency.
  • —Ships deterministic synthetic fixtures with no private coordinates or employer data.
  • —Provides text and JSON CLI output, compressed LAZ support, tests, CI, and explicit limitations.
View the LiDAR systems lab
04

Computer vision collaboration

WEARLYZE

A team-built visual clothing analysis and retrieval project combining data preparation, model training, similarity search, and an interactive segmentation experience.

PythonPyTorchComputer VisionRetrievalSegmentation
  • —Contributed data-loading and transformation pipelines for model-ready inputs.
  • —Built parts of the model, training, and retrieval workflow.
  • —Contributed the garment-segmentation interface used to explore results.
View the team repository
05

Empirical software engineering

AI Development Research

A collaborative study of AI-assisted development, grounded in repository data, reproducible analysis, and careful interpretation of developer outcomes.

PythonGitHub DataStatistical AnalysisResearch
  • —Owned the RQ4 developer-experience analysis and final synthesis.
  • —Built GitHub-enriched pull-request complexity models and audited class imbalance.
  • —Added robustness checks and figures to keep conclusions tied to the available evidence.
Read the research repository

Learning archive

Earlier Work

Smaller projects that document the path to the systems work above.

RoomChatA Flask and Socket.IO real-time chat project.BudgetBuddyA Python price-monitoring and email-alert automation project.BrightDashA Selenium browser-automation project for repetitive portal workflows.TaskFlowAn earlier Flask task-management application.

Let's work together.

I'm interested in systems, product engineering, developer tools, and geospatial work where the implementation and the evidence both matter.