I'm a B.Tech Computer Science & Engineering student at Lovely Professional University, building toward backend engineering, full-stack development, and applied AI. I like the layer underneath a feature: the data model, the API boundary, the rule that prevents an invalid state, and the tests that prove the behaviour.

I'm studying Computer Science & Engineering at Lovely Professional University (2024–2028). My interests are backend systems, full-stack products, databases, problem solving, and practical AI integrations.

I enjoy taking a project beyond a first demo: making the rules explicit, validating inputs, handling edge cases, testing behaviour, and documenting decisions. The projects below are at different stages, so I distinguish implemented functionality from plans and work in progress.

Student opportunity eligibility and assisted-application platform.

EligiCore aims to help students understand which opportunities fit their profiles and move through application preparation with a human checkpoint before final submission. Its design separates resume/profile processing, job ingestion, eligibility, ranking, and application preparation.

• Structured inputs: parse resume information into a reusable candidate profile.
• Explicit decisions: keep eligibility and recommendation outcomes understandable and testable.
• Ingestion boundaries: normalize source records and handle duplicates consistently.
• Human control: do not bypass CAPTCHAs or automatically click final submit on third-party sites.
• Validation-first development: use automated tests to check rules and edge cases as the project evolves.

Stack: Python · API development · data modelling · AI-provider integration · automated tests

Repository: Lakshya172/eligicore ↗

Privacy-aware browser-agent safety layer · project design and early-stage work.

PratiBimb explores how browser automation can handle sensitive information more carefully. The design direction includes typed placeholders for personally identifiable information, explicit verification boundaries, restricted outbound requests, and a memory-only vault concept.

• Privacy by design: avoid casually exposing sensitive values to every automation step.
• Fail-closed verification: stop when a required safety check cannot be satisfied.
• Bounded actions: treat browser actions and outbound requests as policy-controlled operations.
• Honest project status: this should be described as an early-stage design until implementation, tests, and measurements are present in the repository.

Technology direction: TypeScript · browser automation concepts · ONNX Runtime Web · Playwright · Vitest

Repository: [Add the exact PratiBimb repository URL before publishing.]

Full-stack e-commerce platform · Rinky Jewellers, Kanpur.

A commerce application for a real jewellery business, with a customer-facing storefront and an administration workflow for managing products.

• REST backend built with Node.js and Express.
• MongoDB Atlas database with Mongoose.
• Authentication and role-based authorization.
• Product catalogue and admin product management.
• Customer wishlist and order-related workflows.
• Transactional email verification using Brevo.
• Responsive frontend built with HTML, CSS, and vanilla JavaScript.
• Security middleware and request validation.

Stack: Node.js · Express.js · MongoDB Atlas · Mongoose · HTML · CSS · JavaScript · Brevo · Render

Live: rinkyjewellers.in ↗ · Repository: rinky-commerce-engine ↗

A computer-vision project for webcam-based face recognition and attendance recording.

• Webcam-based face detection and recognition.
• KNN-based classification.
• Automated attendance recording and CSV storage.
• Interactive interface for running the workflow.

Stack: Python · OpenCV · KNN · Streamlit · CSV

Repository: Attendence-System ↗

A feature is more than the screen where it appears. I’m learning to reason across the full path: the user action, the application rules, the persistence layer, and the integrations that connect the system to the outside world.

A feature isn't finished just because the happy path works.

A rule should be testable.
A failure state should be deliberate.
Sensitive data should have boundaries.
A project claim should match what the code can demonstrate.

• Make rules explicit. Business logic should be understandable and testable.
• Treat edge cases as design work. Missing data, duplicate records, and invalid states need deliberate behaviour.
• Separate responsibilities. Keep API, persistence, and application logic from becoming one tangled layer.
• Test behaviour, not just execution. A successful response is not enough if the resulting state is wrong.
• Describe work honestly. Distinguish shipped features from experiments, designs, and planned work.

• Backend engineering: service structure, REST API design, validation, and authorization.
• Data structures & algorithms: consistent problem solving and complexity analysis.
• EligiCore: improving ingestion, eligibility, matching, and test coverage.
• Applied AI: understanding useful integrations and evaluating output responsibly.
• Open source: preparing to make meaningful contributions through code, tests, and documentation.
• System design: learning to explain architecture decisions and their trade-offs.

B.Tech — Computer Science & Engineering
Lovely Professional University · 2024–2028

• Data Science & Analytics — HP LIFE
• Effective Communication Skills — Skillera
• Advanced Backend Development — Node.js, Express.js, MongoDB · Grade A, Certificate of Merit · LPU

If you're building practical software, working on backend systems, or interested in thoughtful open-source collaboration, feel free to connect.

GitHub · LinkedIn · Portfolio