I like building software that is technically solid and actually usable. My work spans:
- AI and document intelligence systems
- full-stack product builds with modern web stacks
- practical developer tooling
- open-source contributions in the Kubernetes ecosystem
I care about clean implementation, real-world utility, and shipping projects that do more than look good in a demo.
- Building AI-assisted developer and document workflows
- Going deeper into distributed systems and Kubernetes internals
- Improving product taste on the frontend without sacrificing engineering rigor
Languages Java, Python, C, JavaScript, Go
Frontend React, Vite, HTML, CSS, Tailwind
Backend FastAPI, APIs, auth, storage, RAG flows
AI/ML Gemini, embeddings, OCR, CRF, RAG
Infra GitHub, Docker, Kubernetes, Minikube
- 33 public repositories
- merged PRs in
kubernetes/kubernetes - merged PRs in
kubernetes/minikube - projects across AI, full-stack apps, tooling, and coursework
Real merged work in the Kubernetes ecosystem:
| Repository | Contribution |
|---|---|
kubernetes/kubernetes |
Fixed a CEL composition race by deep-copying MapType state to prevent concurrent map access crashes |
kubernetes/minikube |
Fixed Podman mount behavior on macOS with host.containers.internal support |
kubernetes/minikube |
Corrected nested VM detection on macOS and removed unnecessary timeout inflation |
That combination matters to me because it reflects the kind of work I want to keep doing: debugging real systems, reading unfamiliar code quickly, and landing fixes that improve reliability.
BorderBridge, a humanitarian technology platform for refugee identity verification and case management.
- refugee case workflows for authorities and self-service users
- explainable identity confidence scoring with evidence tracking
- React, TypeScript, FastAPI, Supabase, and custom graph visualisation
Document understanding pipeline for handwritten exam digitization using classical ML.
- CRF-based Q/A segmentation
- OCR + preprocessing pipeline
- practical ML without depending on LLMs
SENTINEL, a hackathon-built mission intelligence console created with a team.
- fused operations view with deterministic scoring and AI reasoning
- React, Vite, Tailwind, FastAPI, and Claude-backed structured analysis
- scenario progression, recommendation paths, and executive briefing flows
Livestock health management platform focused on traceability and bio-safe marketplace transactions.
- digital health passports for animals with treatment history
- deterministic withdrawal-period enforcement before listing products
- Node.js, Express, MongoDB, EJS, Tailwind, and chart-driven dashboards
Core Languages Java, Python, JavaScript, C, Go
Focus Areas AI systems, full-stack apps, open source, developer tooling
Public Work Products, hackathon builds, coursework, Kubernetes contributions
- experiments in AI, OCR, RAG, and developer tools
- full-stack builds with product polish
- Java and DSA practice from fundamentals upward
- open-source work connected to systems engineering
- football
- movies
- finance and economics
- travel
If you're working on AI tools, developer experience, or systems-oriented open source, I'm interested.
- Email: [email protected]
- LinkedIn: Abhigyan Shekhar