Agentic Ecosystem Design: Led the engineering team in the conceptualization and delivery of Spotter as a Model Context Protocol (MCP) Host. This transformation enabled an agentic ecosystem capable of synthesizing unstructured data from Slack, Google Drive, and Loom into actionable business insights.

API & Integration Architecture: Architected and deployed the Spotter MCP Server and public REST API suite. These integrations provide programmatic analytics access and automated data source discovery for external AI clients, bridging the gap between LLMs and enterprise data.

Spotter 3 Delivery: Spearheaded the development of Spotter 3, emphasizing advanced reasoning and thinking capabilities. This allowed the agent to perform complex, multi-step query decomposition and autonomous self-correction using Agent ReAct framework.

High-Precision Learning: Engineered an intelligent skill-based system utilizing few-shot learning and deep learning advancements. This system consistently achieves 90%+ accuracy when processing complex datasets with millions of rows and over 1,000 columns.

Conversational AI Experience: Led the development of a seamless Natural Language Search experience using Java, React, GraphQL, and Python. This included building a custom Data Source Discovery Agent that assists users in identifying relevant datasets through conversational AI, significantly lowering the barrier to entry for data exploration.

Dynamic SQL Inference: Developed LLM-powered ad-hoc formula generation. By leveraging SQL inference, the system handles complex query-on-query operations dynamically, allowing business users to generate sophisticated formulas without technical expertise.

Human-in-the-Loop (HITL): Designed and built a feedback system that empowers users with granular control over database operations, formulas, and chart configurations. This initiative drove customer satisfaction to 95% and significantly improved search precision.

Translated functional and technical requirements into detailed architecture/design/working software solutions by working with PM, UX designers, and customers using Typescript and Javascript along with ReactJS.

Analyzed and optimized UI and infrastructure application code for quality, efficiency, and performance using various tools to improve the performance metrics such as First Paint (FP), First Contentful Paint (FCP), First Meaningful Paint (FMP).

Built a shared component MFE(Micro Frontend) with highly leveraged reusable software platforms, components and APIs for interfacing with various internal applications of different subteams.

Implemented various website performance enhancements on both front-end and backend such as asynchronous loading/operation, DNS prefetching, lazy loading, CDN caching, compression, minification, and asset pipeline with fingerprinting for dynamically generated assets with appropriate content headers.

1) Designed Digital Subscription Service registration system for Tribune Publishing Company’s various newspaper sites that use JavaScript, Ruby on Rails, and Java system to enforce user registration for 8 million users.

2) Created the paywall integration for the Tribune Publishing’s newspapers on Apple news, Facebook instant article, and Google AMP platforms.

3) Implemented the social sign-on (Google, Facebook, Twitter, Yahoo, AOL) registration system for the Tribune Publishing’s user registration system.

4) Developed customer facing and administration web applications using technologies including, but not limited to Ruby on Rails, HTML, DHTML, CSS, JavaScript, and jQuery that is in AWS environment.

5) Implemented CDN caching and fingerprinting for dynamically generated assets from rails application with required headers (Etag, last modified time) and TTL with relevant HTTP status codes (200 and 304).

6) Implemented of the elastic search for content management Rails application for the Tronc, Inc. newspaper sites.

7) Worked on integrating the apple pay for Tribune Publishing web checkout experience.