Eqbal Quran

Created a scalable/fast log management system using a separate MongoDB and migrating the old data from MySQL.

Joined the Dubber team to create new functionalities for the telephony system using WebRTC.

A technology company creating innovative products for real estate professionals and consumers that improves the home search, discovery, and decision-making process.

I contributed to the design and implementation of an MVP for a location-based social application where user-generated updates are visible only to users within specific geographic boundaries. The project combined real-time geolocation logic with LLM-powered personalization to deliver localized feeds, contextual summarization, and relevance-based content ranking.

My work focused on building the core AI architecture, integrating OpenAI APIs for chat and embeddings, and developing a Retrieval-Augmented Generation (RAG) pipeline using LangChain and a vector database to enable semantic search across structured user profiles and unstructured social content. I also implemented geo-contextual data workflows to enforce strict location-based visibility and low-latency delivery.

The solution was deployed on Google Cloud Platform (GCP) as scalable Python microservices, with observability and cost controls integrated via Helicone. This enabled the client to accelerate MVP launch, validate localized user engagement, and establish a scalable foundation for future personalization and in-app digital currency features.

Chromie Dispatch is an AI-powered, SMS-first platform designed to fill last-minute nurse shift gaps in hospitals without requiring EHR or IT integrations. I worked as a Back-End + AI Engineer, leading the design and implementation of the core back-end and intelligent dispatch logic.

I built a production-ready NestJS back end with secure, role-based APIs to support nurse managers, supervisors, and admins. I integrated LLM APIs to power brand-aligned, compliant SMS messaging and decision support, enabling automated outreach, escalation, and confirmation flows. The system included intelligent nurse matching and ranking based on role eligibility, unit fit, availability, response history, and workload constraints, with explainable recommendations for managers.

I also implemented Twilio-based SMS pipelines, scheduling data ingestion from semi-structured CSV/Excel files, audit logging, and operational analytics such as fill rate and time-to-fill. The result was a scalable MVP that allowed hospitals to deploy in hours, automate chaotic manual workflows, and fill shifts in minutes instead of hours.