Priyanshu

Hi, I'm Priyanshu

Developer
View ProjectsContact Me
About Me
Priyanshu
Developer focused on building robust web applications

I'm a Web Developer based in New Brunswick, NJ, focused on building robust in-house web applications as part. At the New Jersey Turnpike Authority, I design and develop custom internal tools, delivering tailored functionality, greater control, and long-term cost efficiency. My work spans full-stack development, API integration, and geographic information systems — ensuring our homegrown applications meet the specific operational needs of the authority. Previously, as a Data Engineer at Kaaye Technologies, I built and optimized data pipelines using Apache Spark, giving me a strong foundation in scalable backend systems that informs how I architect web solutions today.

My research experience includes working as a Research Assistant at Rutgers University, developing classification models for turfgrass species and implementing real-time vehicle speed detection systems. I'm passionate about developing practical AI solutions that bridge theoretical concepts with real-world applications. I've contributed to various ML projects including generative replay models, spiking neural networks, and computer vision systems.

Download Resume
My Projects
Some of my recent work
Generative Replay with Compressed Features
PyTorchBERTGPT-2+1 more

Generative Replay with Compressed Features

Implemented a Hippocampal Memory Indexing module to mimic brain memory storage for NLP tasks, improving task performance by 25% and reducing catastrophic forgetting by 40%.

Evaluating SpikeNet on Temporal Graphs
SnnTorchGraph Neural NetworksTGBN

Evaluating SpikeNet on Temporal Graphs

Benchmarked SpikeNet on a 17M+ edge dynamic graph dataset, achieving 67% accuracy in edge and node property prediction while improving efficiency by 15%.

Real-Time Vehicle Speed Detection
PythonYOLOv7Flask+2 more

Real-Time Vehicle Speed Detection

Hosted a YOLOv7-based live-streaming web app on AWS using Flask, improving image throughput from 12 to 15 FPS with 25% enhanced detection accuracy.

Car Renting Website - REYOCA
Node.jsMySQLAWS+2 more

Car Renting Website - REYOCA

Deployed a car rental website on AWS EC2 with authentication, session management, payment processing via Stripe API, and real-time tracking with Maps API.

See More on GitHub
My Skills

languages

PythonRMATLABC++JavaJavaScriptSQLNoSQLBASH

Machine Learning

PyTorchTensorFlowKERASComputer VisionNLPLSTMSVM

Deep Learning

Reinforcement LearningStableBaseline-3GymnasiumTransformersNeural Networks

Data Engineering

SparkHadoopETL PipelinesDockerKubernetesAWS

frameworks

FlaskNext.jsReactNode.jsSpring BootREST

databases

MySQLMongoDBFirebaseDynamoDBS3
Education

Master of Science in Computer Science

September 2022 - May 2024

Rutgers University, New Brunswick

GPA: 3.92/4.0

Research: Developed classification models for turfgrass species, created vehicle speed detection systems using YOLOv7

Activities and societies: Astronomical Society and Chess Club

Bachlor of Engineering in Electronics and Telecommunication

August 2016 - May 2020

Mumbai University, Mumbai, India

GPA: 3.76/4.0

Robotics Competition 2016: Winner

Activities and societies: Member of Extension Working Committee

Professional Experience

Web Developer

July 2024 - Ongoing

New Jersey Turnpike Authority, Woodbridge, NJ

JavaScriptTypeScriptNextReactAngularGraphQLRESTSOAPPostgres SQLGitJira
  • Developed a web application (FPMS – Fuel Pump Monitoring System) using Next.js and Node.js to display Gasboy station connectivity and transaction data
  • Integrated Loadrite (loader scale system) data—such as salt usage during storm operations—into the NJTA Weather Operations System (NWOS) web app using JavaScript and TypeScript to enable centralized tracking of operational data
  • Integrated the Gasboy web application with the Enterprise Asset Management (EAM) web platform using SOAP API and WSDL to maintain data consistency between applications

Data Engineer

November 2020 - October 2021

Kaaye Technologies LTD, Navi Mumbai, India

PythonDockerSparkTensorFlowMongoDBTableau
  • Partnered with clients — Snapdeal and Dunzo (now Zepto) — to analyze over 2 million customer interactions on new feature rollouts using Tableau and SQL, identifying behavioral trends that improved feature engagement by 15%
  • Built and validated predictive models (accuracy ≥ 85%) to forecast user adoption and retention, enabling stakeholders to make data-driven decisions that optimized product portfolio and reduced underperforming features by 20%

Machine Learning Intern

February 2020 - August 2020

ValueFirst, Gurgaon, India

PythonPyTorchNLTKRASAPyTesseractTensorFlow
  • Automated the employee onboarding process, reducing manual intervention by 35–50% through OCR-based document verification and Python scripting, streamlining data extraction and validation workflows
  • Developed and deployed an AI chatbot using RASA, improving customer interaction efficiency and response time by 40%, while enabling end-to-end query handling through natural language understanding models before GPT 3 was released

Research Assistant

April 2023 - May 2024

Plant Biology Department, New Brunswick, NJ

PythonPyTorchDJI Mavic 3
  • Modelled a VGG-19 classification model for phenotyping 7 different turfgrass species to improve the efficiency and remove a targeted bias of manual phenotyping
  • Generated a Graphical Pedegree dataset for 12 species of turfgrass to obtain a Family link of any plant from any year using retrieval by recursive queries

Research Assistant

November 2023 - May 2024

Durandal Lab, Piscataway, NJ

PythonPyTorchFlaskAWSFFmpeg
  • Implemented a real-time live streaming web application on AWS EC2 using Flask by collecting frames from the PSDK module intrgrated with Yolo object segmentation model
  • Optimized post-detection video processing pipeline, increasing frame rate from 12 FPS to 25 FPS and reducing patchy artifacts, resulting in smoother playback
Publications

Classification of Grains and Quality Analysis using Deep Learning

International Journal of Engineering and Advanced Technology (IJEAT), October 2021

View on Google Scholar
Get In Touch
Have a question or want to work together?
contact@priyanshu.idlinkedin.com/in/priy4nshu