Professional Experience
Collaborated with multiple international clients to architect, optimize, and maintain scalable software solutions across diverse tech stacks.
Key Projects & Clients
Lovelight
- ▹Architected Meilisearch reindexing for 180K+ job activity records by replacing a linear chunked pipeline with an adaptive concurrency dispatcher that scaled parallel workers based on live MySQL connection availability; reduced full reindex time from 40 min to 3.5 min while eliminating DB connection exhaustion and manual reruns.
- ▹Reduced pricing API request payload sizes by 65x (13 MB to 200 KB) and eliminated recurring HTTP 413 production failures by identifying deeply nested frontend object graphs exceeding CloudFront’s 10 MB payload limit and extracting only pricing-critical fields before request serialization.
- ▹Engineered an end-to-end automated Purchase Order ingestion pipeline (Microsoft Graph, S3, PDF parsing) for 250+ builders; implemented builder-specific extraction templates and a side-by-side verification UI, achieving 90% auto-extraction accuracy and reducing processing effort from 10+ minutes of manual entry to a 2-minute review.
- ▹Owned end-to-end design of freight tracking integrations across 4 suppliers (Tomax, Norman, TWC, Ediths), processing 25K+ order updates/day; designed a pluggable strategy-pattern sync layer supporting both webhook and polling flows, reducing duplicate tracking events by 30% and cutting manual shipment-tracking effort by 90%.
- ▹Integrated near real-time Google Calendar and Microsoft Outlook synchronization workflows for 100+ installers handling 150 bookings/day; implemented OAuth 2.0 authentication and polymorphic sync abstractions, reducing API maintenance overhead and eliminating double-bookings.
ReadyTech
- ▹Built a production RAG based Recognition of Prior Learning (RPL) evaluation platform using Amazon Bedrock and Claude Sonnet; ingested 2K+ VET Australian training packages from S3 into pgvector, achieving 87% top-3 competency match accuracy.
- ▹Automated manual assessor workflows by delivering a full-stack portal that evaluates candidate evidence against a custom 5-point rubric, surfacing competency match confidence scores and automating missing evidence gap detection, reducing screening turnaround from 72 hours to under 5 minutes.
Technical Skills
Python
C++
Java
TypeScript
JavaScript
SQL
Data Structures & Algorithms
Overview
View Profile ↗722 submissions in the past year
Total active days: 364
Max streak: 336
Innovative Projects
Pathram - Blockchain-based Document Management and Tracking System
Developed an innovative web-based platform for tracking the status of documents submitted for approval within government departments, focusing on enhancing transparency, security, and efficiency in document workflows.
KEY FEATURES
- ▹Blockchain for Security: Immutable ledger ensures no unauthorized changes.
- ▹Digital Signatures: For enhanced document authenticity.
- ▹Transparency: End-to-end tracking of document approval stages.
- ▹Trust & Tamper-Proofing: Using IPFS for decentralized storage ensures data integrity.
LeetGit
LeetGit is a Chrome extension that automatically syncs LeetCode submissions to GitHub the moment we hit Submit. Every solution is stored as structured Markdown with our code, runtime/memory stats, difficulty, tags, notes, and attempt history. Free, open source (MIT), and no accounts required.
KEY FEATURES
- ▹Automatic capture: Intercepts submissions silently; works with Accepted, Wrong Answer, TLE, and more
- ▹Rich Markdown commits: Each solution file includes runtime & memory percentiles, difficulty, topic tags, and your LeetCode notes
- ▹Duplicate detection: Skips a commit when code and notes haven't changed since the last sync
- ▹Flexible commit messages: Use a template with {number}, {title}, {language}, {status}, {difficulty}, or write a custom message per submission
- ▹Per-problem history table: Tracks every attempt per problem (language, status, runtime, memory, link)
Co-formatter
Format LinkedIn posts, comments, and messages with a native toolbar inside LinkedIn. No copy-pasting or third-party formatting websites required.
KEY FEATURES
- ▹Rich text formatting: Bold, italic, underline, strikethrough, monospace, cursive, uppercase/lowercase, bullet lists, and numbered lists.
- ▹Native LinkedIn toolbar: Built directly into the LinkedIn post editor, comments, and messaging interface. No copy-pasting or external websites.
- ▹Real-time productivity: Live character count, keyboard shortcuts, and Light/Dark mode support for a seamless writing experience.
E-Connect
This chatroom application allows multiple users to exchange messages over Bluetooth or a local Wi-Fi network using WebSocket technology for real-time communication.
KEY FEATURES
- ▹Real-time messaging using WebSockets for instant communication.
- ▹Bluetooth and local Wi-Fi connectivity for seamless chat experiences.
- ▹Clean, responsive UI built with Tailwind CSS for enhanced usability.
- ▹Modular and scalable architecture using Turbo monorepo for optimized development.
My Education
During my B.Tech, I gained a solid foundation in computer science, honing my problem-solving skills through hands-on projects. This experience fueled my passion for software development and taught me the importance of continuous learning in a rapidly evolving tech landscape.
Relevant Coursework
My Publications
A Strategy to Improvise Coin-age Selection in the Proof of Stake Consensus Algorithm
Publisher: IEEE
Date: 10 October 2023
Conference: 2023 International Conference on Software, Telecommunications and Computer Networks (SoftCOM)
Abstract
Proof of stake is one of the consensus protocols in Blockchain technology. Within PoS, there are two common strategies employed for validator selection: Coin-age selection and Randomized block selection. However, both of these strategies have drawbacks when it comes to ensuring a fair selection of validators, leading to a phenomenon known as the 'rich getting richer' syndrome. The paper proposes a new consensus algorithm that introduces a novel parameter: The time stamp of staking tokens. Staking tokens are those minted by validators to participate in the staking process. By considering the timestamp of staking tokens minted before participating in the staking procedure, the proposed algorithm aims to address the shortcomings of the existing strategies. The algorithm seeks to eliminate the loopholes in the coin-age selection method and enhances it in terms of precision. Additionally, it aims to increase the randomness in choosing a validator, thereby mitigating the rich-getting-richer syndrome.
