Problem: E-commerce businesses lacked affordable, integrated AI recommendations and business intelligence tools.

Contribution: Architected and built a production-ready, multi-channel SaaS platform with AI-powered recommendations.

Impact: Delivered a scalable backend, integrated payments, and AI-driven insights for store owners.

Problem: Generic LLMs lacked specific domain context for specialized engineering and operational tasks.

Contribution: Fine-tuned a Mistral-7B model using 137,300 domain-specific training examples across 7 operational modes.

Impact: Built a multi-modal AI assistant supporting engineering, system monitoring, and bilingual interaction.

Problem: Designing and deploying LED animation firmware to multiple microcontrollers was complex and error-prone.

Contribution: Developed a cross-platform desktop application for designing, simulating, and deploying LED animation firmware.

Impact: Supported 29 microcontroller platforms with real-time preview, automation engine, and OTA updates.

Problem: Fragmented automation workflows made discovery and management difficult for enterprises.

Contribution: Built a scalable automation discovery platform indexing 2,000+ workflows and 76,000+ automation nodes.

Impact: Indexed 2,000+ workflows and provided enterprise-grade features including RBAC and analytics.