Who Am I?
I’m a cloud-first AI engineer excelling at building and hosting AI-driven applications on AWS/Azure/Kubernetes. But how did I get here?
It started when I was in sixth grade, where I was introduced to analog electronics. From building circuits on a breadboard with the NE555 timer IC, I evolved into a curious learner, progressing through my tenth grade.
Continuing on this path, my love for the electronics expanded, leading me to explore the domain of embedded systems. I started building simple microcontroller-based hardware projects and ended-up in learning embedded Linux and RTOS. My love for GNU + Linux grew day by day as I started learning a lot more about it. It is this journey that led me to cloud, back-end web development and IoT.
Today, I stand as a seasoned back-end web developer and a skilled agnostic cloud engineer. My expertise spans across Cloud Technologies, IoT, and Embedded Systems, offering a unique fusion of skills that seamlessly merges computer science and electronics, from top to bottom, from low-level hardware to high-level software.
Engineering isn’t just what I do; it’s who I am.
Skills
- Programming Languages (low to high level)
- C (compiled)
- Go (compiled, garbage-collected)
- Python (scripting)
- AI AI AI (and ML)
- Agentic Systems (includes RAGs)
- MCPs
- Basic Supervised Learning
- Back-end Web Development
- Flask, Django and Django REST
- Databases
- MySQL & PostgreSQL
- MongoDB
- Cloud (agnostic)
- AWS 📜
- Azure 📜
- Docker & Kubernetes (EKS and AKS)
- Ansible
- HashiCorp Terraform and Packer
- Embedded (µC/µP)
- Atmel AVR
- ESP82xx
- Raspberry Pi
- Raspberry Pi Pico
- Zephyr RTOS
- Miscellaneous
- Minimalist
- Zero Plugin Neovim/Vim User
Certifications
Projects
Deep Researcher
The "Deep Researcher" project offers an agentic workflow that understands your query, generates queries, searches the internet and writes a detailed report for you. This framework allows users to integrate various crawlers and foundational model providers to enhance their research capabilities, making it a flexible and extensible tool for AI-powered research. It breaks down your query into multiple queries (breath-wise learning, researching related stuff) and for each query, it generates learnings. With those learnings, it asks further questions (depth-wise learning learning, providing richer research). This makes it a very performant and expert researcher.
AI Enhanced CCTV Surveillance
The objective of this project is to introduce intelligence to CCTV systems. Raspberry Pi based edge-devices are placed in places of interest. These edge-devices run a custom binary image classification model built upon the TFLite Runtime library. The image captured by a camera attached to the end-device is inferred by the model and the prediction results are uploaded to a website. The website is built on Flask WSGI. It provides a dashboard to monitor images uploaded by various edge-devices and add/remove/update users and devices. Users that are added will be informed via SMS and email the information (image, location, time and date) of theft immediately. Also, only the edge-devices that are registered with the website can upload data. A end-device is assigned a UUID when it is registered with the website to provide a layer of authentication. This solution is readily deployable and manageable at a large scale.
Glance Village Hackathon-2023: Wildlife Interference Monitoring System
This project focuses on reducing the number of deaths caused due to wildlife interference. Raspberry Pi based edge-devices are placed in forest areas around the village with camera that infer images with a custom multi-class CNN image classification model built upon the Tensorflow library. All the data is uploaded to a centralized Flask WSGI server. The forest department can access a dashboard that streams live image feeds from all the edge-devices, add/remove/update first responders (to whom messages will be sent during calamity), view and download reports. If any image is classified as containing a wild animal, the Flask server automatically alerts all the first responders (villagers and forest officers) via SMS so that necessary actions can be carried out by them.
Tensorflow Custom Object Detection
A Tensorflow object detection model based on the SSD MobileNet built for Smart India Hackathon-2022 as finalists.
Tensorflow Sentiment Analyser
A LSTM sentiment analyser that can analyse sentences based on six categories.
About This Website
This website is a single page application stylized using Bootstrap. Not a lot of effort has been put onto the front-end side, except for its responsiveness and consistency in design.
The website is served by an Nginx server running on AWS EC2. This project can be containerized if needed. The project also is completely automated with a CI/CD pipeline that is built using AWS's services. Provisioning of these various services is automated with the IaC tool called Terraform. This may sound a bit over-engineering, but I call it learning.
Social Profiles/Contact Me
