Hi, I’m Shrey. I’m a computer engineer currently leading engineering for Quantum’s Unified Surveillance Platform (USP), a hyperconverged infrastructure solution designed for large-scale video surveillance. Alongside my day job, I actively explore applied AI, particularly where it intersects with distributed systems and developer productivity.
Before Quantum, I was a founding engineer at EnCloudEn, where I helped build virtual desktop, private cloud, and HCI platforms from the ground up. I joined as the second engineer, grew into the Director of Engineering role, and led a 15-member team whose work ultimately formed the foundation for EnCloudEn’s acquisition by Quantum.
I graduated from NIT Trichy in 2014 with a degree in Computer Science. During college, I served as president of the dance team — an experience that shaped many of my leadership instincts. Leading through wins, losses, and high-pressure moments taught me resilience, accountability, and how to bring people together toward a common goal.
When I’m not building or thinking about systems, you’ll usually find me chasing some sport or the other.
AI is the biggest buzzword in tech today, and like many engineers, I've been diving deep into it. Since my day-to-day work doesn't revolve around AI, I began taking on side projects to learn, experiment, and document my progress. I genuinely believe that AI is something every software engineer should explore, because it marks a major paradigm shift in how we build software—We've moved from machine code to assembly to high-level languages… and now to AI-assisted development. More on AI
A lightweight AI-powered tool that converts raw business requirements into structured user stories, acceptance criteria, test cases, and functional summaries. Supports text/PDF/DOCX input, and provides the generated output in a downloadable DOC file.
Technologies: Python, Streamlit, OpenAI API, python-docx, PyPDF2 / python-docx
Built an AI-powered sales analytics tool that automatically generates interactive charts, KPIs, and business insights from a CSV dataset. The system analyzes sales trends, category/region performance, and produces executive summaries and recommendations using an LLM
Technologies: Python, Streamlit, Pandas, Plotly, Matplotlib, OpenAI API
Zerodha MCP Server & Client — An AI-powered trading assistant that lets you manage your Zerodha trading account through natural language. Built with Python, the Model Context Protocol (MCP), and the Agno framework, it supports placing orders, checking positions, viewing portfolio holdings, and managing trades via a conversational interface or web UI
Technologies: Python, Agno Framework, MCP, KiteConnect, OpenAI, Gradio
An AI-powered hotel recommendation system that scrapes Google and Booking.com, then uses GPT-4o-mini to analyze reviews and score hotels by customizable criteria (breakfast, cleanliness, service, location, value). Provides ranked recommendations with insights, strengths, and weaknesses to help users make informed booking decisions. Features a web interface with real-time streaming results.
Technologies: Python, Flask, LangGraph, Bright Data MCP, OpenAI, React
Open Source AI Debugging Platform
Creator & Maintainer
While building the backend for my startup, production bugs were often discovered only after customer reports due to limited observability, leading to slow and reactive debugging.
I started with a lightweight mechanism to capture and persist stack traces on runtime errors, prioritizing fast feedback and minimal integration overhead for an early-stage product. This evolved into an error ingestion and debugging platform with structured storage and filtered retrieval of error events. On top of this foundation, I integrated an AI-driven analysis pipeline that performs first-level debugging using stack traces and selectively fetched source code from Git.
StackWise is an open-source, self-hosted debugging tool focused on helping small teams and individual developers reason about production errors more effectively. Rather than replacing full observability stacks, it complements them by reducing the manual effort required to understand and triage recurring failures.
The project is currently maintained and developed independently.
StackWise is designed with extensibility in mind. Possible future directions include:
These ideas build on the existing ingestion and analysis foundation while keeping the system developer-centric and self-hosted.