<Hello World />
Aditya Jhaveri

Aditya Jhaveri

Software Engineer | AI/ML Enthusiast | Distributed Systems Developer

Every Computer Science journey begins with Hello World. Mine began with Age of Empires, learning strategies and problem-solving before writing my first line of code. Today, I’m a Computer Science graduate student at NYU Tandon with 3+ years of experience building scalable systems and leveraging AI to solve complex problems. I specialize in distributed systems, big data processing, and full-stack development.

Get In Touch View GitHub

< About Me />

3+ Years

of Professional Software Engineering Experience

3B+
Records Managed via HDFS
5+
Microservices Integrated with Kafka
30%
Faster Data Extraction
20%+
System Performance Gain @ NYU

< Professional Experience />

Software Developer

New York University
Feb 2025 - Present
Internal SaaS / Ops

Developing workflow software for NYU's Global Enrollment Management and Student Success team, supporting enrollment operations, staff communications, and service processes used by campus administrators.

Role Focus

Partner with EM tech admins to design and improve Google Cloud-based SaaS solutions across the full lifecycle: backend logic analysis, front-end development, documentation, alpha testing, and production optimization.

Impact Highlights

  • Migrated production Google Apps Script tools toward a Next.js + MongoDB architecture, enabling real-time usage and a consistent 60-second sync cycle
  • Engineered a fault-tolerant, concurrency-safe form delivery system, reducing retry failures by 40% during peak submissions
  • Built and owned a placeholder-driven email template platform that auto-populates user, credential, and device data, reducing email management effort by 55%
  • Implemented compound label-based filtering (UI + backend query logic), reducing lookup time across 2,000+ records
  • Improved Google Apps Script retrieval speed by 50% by replacing service-layer calls with direct Google Sheets API access
  • Architected a PoC migrating 1,000+ Google Sheets records to Firestore and Google Cloud SQL for scalability and reliability
  • Built an automated Shipping Label Page that reduced manual processing time by 50%

Stack & Skills

JavaScript Google Apps Script Google Sheets API Next.js MongoDB HTML UI/UX Testing Technical Documentation

Software Development Engineer

Sainapse
July 2022 - May 2024
Data Platforms

Worked in Sainapse's Research, Technology and Platform department, building platform capabilities that supported large-scale data ingestion, transfer, and downstream analytics and ML workflows.

Role Focus

Developed PoCs, solved complex production bugs, and implemented performance-focused platform improvements across microservices, storage systems, and data transfer pipelines on AWS (Linux).

Impact Highlights

  • Optimized microservice file transfers via Apache Kafka using byte-level serialization/deserialization and parallel file distribution, reducing time complexity by 50% and space complexity by 33%
  • Spearheaded HDFS implementation in Java for high-volume ingestion, storing 3B+ daily records with Hive across 10+ microservices and improving average system performance by 37%
  • Designed a batch data adapter attached to BigQuery tables, enabling on-demand transfer of 2B+ rows for downstream ML workflows including free-text search and deduplication
  • Improved extraction speed from 3GB+ XLSX and DOCX files by 30% by updating parsing logic with the Apryse Java SDK
  • Pioneered an Apache Thrift proof of concept for cross-language model development across Java, Python, and Scala
  • Mentored 4 university interns on product architecture and core technology stack, enabling independent feature delivery

Stack & Skills

Java Apache Kafka HDFS Apache Hive AWS (Linux) BigQuery Apryse SDK Apache Thrift Python Scala Data Pipelines PoCs Mentorship

Data Science Intern

AiDash
Jan 2022 - June 2022
GeoAI / Remote Sensing

Supported geospatial ML work at AiDash by developing vegetation classification methods and LANDSAT image classification workflows for remote-sensing use cases.

Role Focus

Built statistical approaches for vegetation classification, trained and evaluated LANDSAT machine learning models, and assessed model quality using metrics such as accuracy, precision, recall, and F1 score.

Impact Highlights

  • Evaluated ResNet architectures with Keras and TensorFlow for LANDSAT classification, achieving 60%-75% accuracy across experiments
  • Devised a Python-based grassland classification method to improve land ranking, achieving 78% accuracy and ranking 1st among peer solutions
  • Built a Python script to automate satellite image labeling in QGIS, reducing manual processing time by 80%
  • Resolved geospatial data handling during onboarding by performing CRUD operations on 10,000+ shapefiles using GeoPandas and PostgreSQL

Stack & Skills

Python TensorFlow Keras ResNet LANDSAT QGIS GeoPandas PostgreSQL Geospatial ML Model Evaluation Precision/Recall/F1 Data Labeling

Data Analyst Intern

PayPal
May 2020 - June 2020
Forecasting / Analytics

Supported demand analytics at PayPal by developing and maintaining forecasting models to identify trends in customer demand behavior.

Role Focus

Built forecasting models, evaluated regression performance across multiple loss functions, and translated findings into a clear presentation with a peer team.

Impact Highlights

  • Identified 10+ customer demand trends and developed forecasting models with R² values close to 0.9
  • Achieved R² scores of 0.98, 0.91, and 0.85 for squared, absolute, and infinite loss functions in a linear regression problem, demonstrating robustness to outliers
  • Collaborated with a team of 4 peers to deliver a comprehensive presentation on forecasting customer demand

Stack & Skills

Python Forecasting Linear Regression Regression Loss Functions R² Evaluation Model Evaluation Data Analysis Presentation

< Featured Projects />

Market Data Service

Assessment Demo / Microservices

A company assessment project demonstrating an event-driven market data microservice that polls stock prices, computes moving averages, and serves analytics-ready data through FastAPI.

What It Solves

Shows how to build a production-style backend for real-time market ingestion and derived analytics, with separation between API handling, async event processing, caching, persistence, and provider integration.

What I Built

  • Built a FastAPI microservice for polling stock prices, computing moving averages, and exposing REST endpoints for latest prices and polling jobs
  • Integrated Finnhub for live market quotes, Kafka for event publishing/consumption, Redis for caching latest prices, and PostgreSQL for raw + processed data storage
  • Designed a Docker / Docker Compose local stack for service orchestration and reproducible setup
  • Added SlowAPI-based rate limiting, pytest coverage for sync/async routes, and GitHub Actions CI for automated testing and Docker build validation

Impact / Results

  • ~40% improvement in data processing scalability
  • ~25% reduction in average API response time
  • Delivered a clean, testable event-driven architecture suitable for assessment/demo review

Stack & Skills

FastAPI PostgreSQL Redis Apache Kafka Finnhub API Docker Pytest SlowAPI GitHub Actions Event-Driven Architecture
View Code

AI Project Suite

Coursework / Multi-Assignment

A collection of Artificial Intelligence course assignments built to implement core topics from class across theory, machine learning, computer vision, planning, retrieval, and data pipelines.

What It Solves

Demonstrates end-to-end application of AI concepts beyond textbook exercises by turning lecture topics into working implementations: statistical modeling, anomaly detection, object detection/tracking, planning-based LLM routing, and multimodal retrieval pipelines.

What I Built

  • Implemented assignments covering Information Theory, Exponential Distribution, polynomial linear regression, and Stochastic Gradient Descent with explanations documented in notebook markdown blocks
  • Built anomaly detection workflows using the anomalib library with PatchCore and EfficientAD, along with related receptive field analysis and similarity-search components
  • Developed computer vision tasks for object detection (YOLOv4-CSP), Kalman filter-based tracking, and vehicular traffic counting from video streams
  • Designed a PDDL-based LLM routing system (domain + problem + validator/solver flow) and paired it with an OpenRouter API implementation for practical runtime request routing
  • Built a video-data ETL and featurization pipeline using MongoDB, OpenCLIP, BERT, spaCy, and Qdrant to index lecture video frames and support semantic retrieval with timestamp-based video clipping

Impact / Results

  • Translated multiple AI course topics into working systems and notebooks spanning theory-heavy and systems-oriented implementations
  • Demonstrated breadth across planning, CV, anomaly detection, retrieval, and ETL/feature pipelines in one consolidated project suite
  • Created reusable notebook-based implementations with documented explanations for coursework review and technical discussion

Stack & Skills

Python Jupyter TensorFlow anomalib PatchCore EfficientAD Qdrant OpenCLIP BERT spaCy MongoDB PDDL Fast Downward OpenRouter API YOLOv4-CSP Kalman Filter
View Code

Apple Wallet Coupon System

NYU / Workflow Automation

Built during my NYU GEMSS role, this system automates Apple Wallet coupon campaign creation, distribution, and redemption using Google Apps Script, PassKit, and Google Sheets.

What It Solves

Replaces manual coupon campaign setup and redemption tracking with an admin-friendly workflow that supports campaign creation, multi-channel distribution, QR-based redemption, and automated expiry notifications.

What I Built

  • Built a Google Apps Script application to create Apple Wallet coupon campaigns with configurable service types and campaign details
  • Implemented coupon distribution through QR code, email (GAS), and SMS (Twilio) channels
  • Integrated PassKit API to create and manage Apple Wallet passes and support redemption workflows
  • Designed QR scanner-assisted redemption flows with validation and fallback handling for coupon lookup in Google Sheets / PassKit
  • Used Google Sheets for campaign and coupon record management, and added a self-triggering script for expiry notifications
  • Built a responsive UI with HTML/CSS and Bootstrap for easier admin operations

Impact / Results

  • Generated 500+ Apple Wallet coupons through the automated workflow
  • Improved operational efficiency by ~40%
  • Reduced manual handling across campaign creation, distribution, and redemption tracking

Stack & Skills

Google Apps Script JavaScript HTML/CSS Bootstrap Google Sheets PassKit API Twilio QR Workflows Automation
View Code

Computer Vision Portfolio

Coursework / CV Portfolio

A consolidated portfolio for CS-GY 6643 (Computer Vision) that packages multiple course projects spanning classical image processing, segmentation, detection/tracking, multimodal Kaggle workflows, and geolocation.

What It Solves

Creates a single organized repository for diverse computer vision assignments and experiments, making it easier to demonstrate breadth across classical CV, deep learning, tracking, medical imaging, and Kaggle-style production workflows.

What I Built

  • Project 01: Built a classical vision pipeline for astronomical image restoration, multi-scale template matching, and geometric validation (including Procrustes alignment)
  • Project 2: Developed a multi-organ nuclei segmentation and classification pipeline with CellPose fine-tuning, XML polygon parsing, Albumentations transforms, and Kaggle-ready RLE submission generation
  • Project 3: Implemented coursework and Kaggle workflows for object detection, Kalman filter tracking, and baseball pitch-tracking using multimodal (vision + tabular) approaches
  • Project 4: Built a GeoGuessr-style U.S. state geolocation pipeline using transfer learning on torchvision backbones with train/resume/infer modes and checkpointed submissions
  • Maintained structured project layouts, local environment reproducibility notes, and assignment write-ups to support onboarding and repeatable experimentation

Impact / Results

  • Demonstrates end-to-end coverage of the CV curriculum across restoration, segmentation, tracking, multimodal modeling, and geo-localization
  • Combines notebook-based experimentation with reusable scripts, checkpoints, and submission workflows for practical reproducibility
  • Showcases ability to move from assignment prompts to working pipelines across both classical and deep learning computer vision tasks

Stack & Skills

Python PyTorch OpenCV Albumentations CellPose YOLO Kalman Filter Jupyter Kaggle Workflows Transfer Learning Geospatial CV HPC Training
View Code

< Technical Skills />

Languages

Core
Java Python JavaScript C++ SQL Bash HTML5/CSS

Frameworks & Tools

Build & App
Spring Boot React Maven Docker Git AWS GCP Kafka Hadoop Spark

AI / ML Libraries

Modeling
TensorFlow Keras NumPy Pandas SciPy Qdrant OpenCV

Databases

Storage
PostgreSQL MySQL MongoDB BigQuery Apache Hive/Pig JPA Hibernate

Dev Tools

Workflow
Postman API Thrift IntelliJ Jupyter Notebook PyCharm JIRA VS Code QGIS

< Education />

Master of Science in Computer Science

New York University - Tandon School of Engineering

Sep 2024 - May 2026
Graduate
NYU Tandon Computer Science New York, USA

Coursework focus: Algorithms, Big Data, Options Pricing and Stochastic Calculus, Search Engines.

Bachelor of Engineering in Computer Science

Birla Institute of Technology and Science, Pilani

Aug 2018 - July 2022 | Minor in Data Science
Undergraduate
BITS Pilani Computer Science Minor: Data Science

Coursework focus: Data Mining, Deep Learning, Machine Learning, Natural Language Processing.

< Let's Connect! />

I'm always interested in collaborating on projects involving AI/ML, distributed systems, or big data. Feel free to reach out!

[@] [email protected]
[>>] LinkedIn
[</>] GitHub
[*] Jersey City, NJ

< Resume />

Preview my latest experience right here on the site or grab a copy to review later.

open docs/resume.pdf · updated 2025

Preview unavailable in this browser.

Download PDF
Download PDF Open in New Tab

Contact Me

Have a question or want to collaborate? Drop a message below!

Rocket