Built the Competitive Intelligence platform from zero to a live, customer-facing product: an autonomous Gen AI pipeline that turns millions of documents (meeting recordings, salary data, Slack, Notion) into structured company dossiers, with a two-pass facts-and-memory design that eliminates LLM hallucination on critical fields. Now core to the enterprise offering and directly contributing to closing enterprise deals.
Exposed the platform over the Model Context Protocol (MCP), giving internal teams deep semantic search across all competitive data from Claude, ChatGPT and Cursor.
Hardened the backend against scraping and denial-of-service attacks, including one that spiked load from 800 to 19,000 requests per minute, by engineering a load-shedding and circuit-breaking layer that sheds traffic before overload and preserves server capacity.
Built a salary aggregation engine powering Industry Pages, serving precomputed medians and percentiles over 10M+ compensation records in under 50ms.
Led the PostgreSQL 12.9 to 17.4 upgrade across all services with zero data loss and near-zero downtime, and redesigned backup and retention to cut ~$18,000/year in cost while staying SOC 2 compliant.
Raised team-wide engineering leverage by building an in-house Claude-powered agent that turns GitLab issues into ready-to-review merge requests, and by driving adoption of AI-assisted code review across the team.
Led end-to-end planning, design, and technical implementation of workspace restriction feature, ensuring compliance with enterprise needs, which resulted in increased ARR by $4 Million.
Proactively addressed system impact by implementing NewRelic alerts, reducing service degradation incidents by 50% and maintaining high standards of observability.
Improved performance on the All Workspaces Page for larger teams by reducing the time to first interaction from ~26 seconds to ~0.6 seconds.
Led the release of the Data Governance initiative. Implemented an asynchronous workflow using Temporal that facilitated controlled migration for large enterprise teams without overwhelming any internal systems, ensuring stability and seamless
transition within 7 hours for 600+ enterprise teams.
Improved service simplicity through the redesign of the archival layer for the Chronicle service (which powers the Activity feed ↗️ in Postman), resulting in better control over the archival process and reduced dependencies.
Streamlined workspace creation process with Workspace Templates, resulting in 60% increase in usage of templates Week-over-Week.
Developed and deployed standalone UI components to NPM, enabling cross-team code reuse and enhancing development efficiency.
Optimized search performance for products by implementing AWS CloudSearch, resulting in a remarkable average response time reduction from 2.5 seconds to 0.4 seconds. Led the end-to-end implementation, including proof of concept (POC) validation and seamless data migration using Lambda functions.
Transformed a legacy system from JavaScript to TypeScript, significantly improving maintainability and code quality. Introduced Lambda architecture to the team, leveraging serverless computing for services where it provided optimal performance and scalability.
Enhanced system observability by implementing proactive failure detection alarms. This initiative expedited the resolution of critical production issues, minimizing downtime and ensuring smooth operation.
Enhanced Admin and Super Admin flows, incorporating features that facilitated upselling existing features for Admins and resulting in increased revenue.
Developed and released a major version upgrade for the Newsletter system, significantly improving performance and resolving bugs.
Built a web data API for LLMs and agents that turns any URL into model-ready content, collapsing a 59,769-token page of HTML into 6,831 tokens of clean Markdown, with scraping, crawling, site mapping, search and schema-based extraction all returning the same response shape.
Set an open extraction benchmark over 1,497 human-labelled pages across 1,296 domains, scoring the engine against six other extractors on identical pages and labels. It lands at 82.9 word-level F1, ahead of every widely deployed open extractor, and the whole harness is published MIT-licensed on GitHub so any number on it can be reproduced in one command.
Used that corpus to show what article-only benchmarks hide: the field is separated by 10.9 points on articles but 51.1 on the service and product pages agents actually read, and published the losing column alongside the winning one rather than only the flattering half.
Engineered the system to scale from a single URL to an entire site: JavaScript rendering, challenge walls, proxies and retries are absorbed server-side, long crawls run as asynchronous jobs with webhooks, and a whole site’s inventory comes back for one credit.
Made the economics as predictable as the API, billing one credit per delivered page with no surcharge for whatever it took and nothing charged for a page that failed, so cost is a function of results rather than of effort.
Shipped it as a product surface teams can adopt in minutes: official SDKs for Node, Python, Go, Ruby, Rust and PHP, plus an MCP server that lets an agent reach the live web mid-conversation.
Built an intent layer for B2B outbound that surfaces buyers showing real buying signals, rather than selling static contact lists.
Detects intent signals across job changes, funding announcements, hiring activity, competitor engagement and category discussions on LinkedIn, refreshed daily.
Designed a warmth scoring model that ranks leads on signal strength, ICP fit and recency, so the output is a prioritised queue instead of a contact dump.
Generates pre-drafted openers that reference the specific “why now” signal, cutting the manual research step out of outbound.
Shipped supporting surfaces including a browser extension for LinkedIn outreach, CRM push, and multi-product workspaces for agencies.
Built a comprehensive digital estate planning platform addressing the $200+ billion in digital assets lost annually, helping families avoid the typical 6+ month search for critical documents during grief.
Architected a zero-knowledge security system with client-side AES-256 encryption ensuring no plaintext data ever touches the server, providing bank-level security for sensitive documents, passwords, and cryptocurrency keys.
Designed an intelligent heartbeat monitoring system with automated daily check-ins and multi-channel family alerts (email, Discord, Telegram) to handle user incapacitation scenarios.
Implemented Shamir’s Secret Sharing algorithm with configurable trust levels for trusted contacts, enabling secure vault recovery while maintaining cryptographic integrity.
Developed end-to-end full-stack solution including billing system, real-time notifications, transparent activity logging, and planned integrations with password managers and cloud storage providers.
Developed a high-performance CLI tool in Go for automated Git repository backup and synchronization across multiple hosting platforms (GitHub, GitLab, Bitbucket, Gitea, Forgejo).
Implemented concurrent processing architecture enabling simultaneous backup of multiple repositories with configurable clone strategies (bare, shallow, mirror, full).
Achieved significant community adoption with 500+ GitHub stars and 18,000+ Docker downloads, addressing critical developer need for repository data resilience.
Designed YAML-based configuration system with customizable backup directories and periodic synchronization, providing proactive solution against account bans and platform issues.
Built with minimal storage optimization and cross-platform compatibility, ensuring efficient backup processes for developers managing multiple repositories.
Developed a static site showcasing a satirical yet insightful list of professional alternatives for corporate communication.
Implemented features such as a fuzzy search on the content, favoriting options, and PDF printing for improved user experience.
Achieved exceptional traction with 100K+ visitors within 24 hours of launch and currently registered 500K+ visitors globally, generating buzz on Hackernews and Reddit.
Scaled infrastructure by migrating to static export and migrating to Cloudflare to leverage the enhanced CDN support and cache control.
Education
Vidyavardhini’s College Of Engineering & Technology
Bachelor of Engineering in Information Technology | June 2016 - June 2020
Graduated with CGPA of 8.54 / 10.
Honors & Awards
[2020] Top 5 Finalist / 1200 Students, Smart India Hackathon 2020
