Pranav Hirani

Building Scalable AI & Backend Solutions | Cloud & DevOps Expert

Developed AI-powered NLP services leveraging Microsoft Phi-3 LLM, FastAPI, and Python to automate form autofill & system data extraction for enterprise applications.

Architected large-scale e-commerce scrapers for Amazon & Flipkart using Scrapy & PostgreSQL, implementing automated job scheduling for real-time data updates.

Designed & deployed microservices-based .NET Web API applications using the Aspire framework, with multi-environment CI/CD pipelines utilizing GitHub Actions, and Bicep (Azure IaC) in Azure Container Apps.

Single-handedly managed cloud infrastructure using Terraform, provisioning, securing, and optimizing cloud environments for high availability & cost efficiency via a modular configuration approach for the current application architecture, making it fully customizable via variables.

Architected a fully event-driven system for search and analysis of SaaS applications using Azure services, enabling scalable and real-time data processing.

Designed and created ETL pipelines using Azure services to streamline data ingestion, transformation, and analysis for data-driven insights.

Implemented fine-grained authorization using OpenFGA and Topaz to enable ReBAC in Aspire framework-based microservices .NET Web API applications, enhancing security and access control.

Successfully managed and executed migration processes to upgrade internal SaaS application versions, ensuring seamless transitions and minimal downtime.

Building Scalable AI & Backend Solutions | Cloud & DevOps Expert Developed AI-powered NLP services leveraging Microsoft Phi-3 LLM, FastAPI, and Python to automate form autofill & system data extraction for enterprise applications. Architected large-scale e-commerce scrapers for Amazon & Flipkart using Scrapy & PostgreSQL, implementing automated job scheduling for real-time data updates. Designed & deployed microservices-based .NET Web API applications using the Aspire framework, with multi-environment CI/CD pipelines utilizing GitHub Actions, and Bicep (Azure IaC) in Azure Container Apps. Single-handedly managed cloud infrastructure using Terraform, provisioning, securing, and optimizing cloud environments for high availability & cost efficiency via a modular configuration approach for the current application architecture, making it fully customizable via variables. Architected a fully event-driven system for search and analysis of SaaS applications using Azure services, enabling scalable and real-time data processing. Designed and created ETL pipelines using Azure services to streamline data ingestion, transformation, and analysis for data-driven insights. Implemented fine-grained authorization using OpenFGA and Topaz to enable ReBAC in Aspire framework-based microservices .NET Web API applications, enhancing security and access control. Successfully managed and executed migration processes to upgrade internal SaaS application versions, ensuring seamless transitions and minimal downtime.

Skills: DevOps Microsoft Azure Product Development .NET Framework Odoo

AI-Powered Image Processing | Cloud & Kubernetes Specialist

Spearheaded the deployment of a Java Spring Boot application integrated with InfluxDB on Kubernetes, optimizing performance through custom Helm charts.

Developed advanced Computer Vision pipelines for wall segmentation, ambient occlusion, perspective matching, and object dimension estimation, leveraging Hugging Face models.

Optimized AI-powered image processing workflows, improving accuracy, efficiency, and automation in large-scale image analysis.

Integrated cloud-based AI models into production systems, ensuring scalability, security, and high availability.

AI-Powered Image Processing | Cloud & Kubernetes Specialist Spearheaded the deployment of a Java Spring Boot application integrated with InfluxDB on Kubernetes, optimizing performance through custom Helm charts. Developed advanced Computer Vision pipelines for wall segmentation, ambient occlusion, perspective matching, and object dimension estimation, leveraging Hugging Face models. Optimized AI-powered image processing workflows, improving accuracy, efficiency, and automation in large-scale image analysis. Integrated cloud-based AI models into production systems, ensuring scalability, security, and high availability.

Skills: Kubernetes Computer Vision Docker Docker Swarm

AI-Driven Computer Vision | Real-Time Tracking & ASR

Designed & implemented a real-time people tracking system using YOLOv8 to analyze restaurant occupancy & foot traffic, improving detection accuracy by 40%.

Developed a custom object detection system using YOLOv5 & YOLOv8, improving precision through dataset fine-tuning & model optimization.

Built an end-to-end Automatic Speech Recognition (ASR) system using OpenAI Whisper LLM, enabling real-time audio transcription.

Engineered high-performance AI inference pipelines, optimizing GPU acceleration (TensorRT) on NVIDIA A6000, Tesla T4, and 4090TI, boosting FPS by 50%.

Deployed custom Flask-based APIs for seamless integration of AI models into production systems, improving response time & scalability.

Implemented an OCR-powered restaurant menu parser using easyOCR, automating menu item & price extraction.

Developed a manufacturing line tracking system using YOLOv8, BotSORT & OC-SORT, accurately counting & tracking engine components.

AI-Driven Computer Vision | Real-Time Tracking & ASR Designed & implemented a real-time people tracking system using YOLOv8 to analyze restaurant occupancy & foot traffic, improving detection accuracy by 40%. Developed a custom object detection system using YOLOv5 & YOLOv8, improving precision through dataset fine-tuning & model optimization. Built an end-to-end Automatic Speech Recognition (ASR) system using OpenAI Whisper LLM, enabling real-time audio transcription. Engineered high-performance AI inference pipelines, optimizing GPU acceleration (TensorRT) on NVIDIA A6000, Tesla T4, and 4090TI, boosting FPS by 50%. Deployed custom Flask-based APIs for seamless integration of AI models into production systems, improving response time & scalability. Implemented an OCR-powered restaurant menu parser using easyOCR, automating menu item & price extraction. Developed a manufacturing line tracking system using YOLOv8, BotSORT & OC-SORT, accurately counting & tracking engine components.

Skills: Machine Learning MLOps Amazon Web Services (AWS) Computer Vision Containerization

A small internship related to GTU curriculum.

In this internship I learned about how to analyze data, worked on APIs, data serialization using pandas, visualization of meaningful combinations of data and done forecasting using regression techniques.