Lately, I've been reaching out to professionals across the industry through cold emails. If you're one of the people I contacted and you've landed here through the link in my email signature — thank you! I truly appreciate you taking the time to visit. It means a lot to me.

Jayraj Pamnani

M.S. Computer Engineering, New York University

Tandon School of Engineering · May 2026

Jayraj Pamnani
jmp10051@nyu.edu · LinkedIn · GitHub · Resume

About

I recently graduated with a Master’s in Computer Engineering from New York University’s Tandon School of Engineering. My work sits at the intersection of machine learning, deep learning, and systems engineering, with a focus on building practical, efficient, and scalable AI systems.

I have experience developing and optimizing AI pipelines across model training, evaluation, deployment, and MLOps. My interests include backend engineering, AI infrastructure, cloud-native systems, and making large-scale models more reliable and production-ready.

Prior to NYU, I completed my B.Tech. in Computer Science & Engineering with a specialization in Artificial Intelligence at Parul University in India, where I built a strong foundation in data structures, pattern recognition, machine learning, and GPU computing.

Education

M.S. Computer Engineering — New York University, Tandon School of Engineering May 2026

Coursework: High Performance Machine Learning, Machine Learning Operations, Deep Learning, Database Systems, Big Data.

B.Tech. Computer Science & Engineering (AI Specialization) — Parul University, India May 2024

Coursework: Data Structures & Algorithms, Machine Learning, Deep Learning with NLP, Pattern Recognition, Image Processing, GPU Computing, Data Visualization.

Technical Skills

Languages
Python, SQL, C/C++, Java, JavaScript
ML/AI Frameworks
TensorFlow, PyTorch, Matplotlib, Seaborn, Scikit-learn, Neural Networks, Computer Vision, NLP
Data & Cloud
Distributed Systems, AWS, CI/CD, MongoDB, PostgreSQL, Hadoop, Spark, ETL Pipelines, Tableau Dashboards
Tools
Git, Docker, Kubernetes, Jupyter, Gradio, Hugging Face, WandB, n8n, Postman, LangChain

Professional Experience

Software Engineering Intern — GBCS Group, Calgary, Canada Nov 2025 – Feb 2026
  • Implemented and optimized cloud infrastructure using Infrastructure as Code (IaC) to improve resource utilization, resulting in a 28% reduction in hosting and maintenance costs.
  • Improved CI/CD pipelines and deployment workflows, accelerating release cycles by 40% while maintaining 99.9% system availability.
  • Helped redesign microservice boundaries and deployment workflows across a 5-engineer team, reducing technical debt, improving maintainability by 30%, and establishing patterns for future service development.
  • Defined deployment standards for Docker/Kubernetes-based services, improving release reliability and reducing manual intervention.
Freelance Software Developer — New York, NY Jul 2025 – Sep 2025
  • Engineered a Django-based web app automating data transfer between QuickBooks and KatanaMRP, cutting manual bookkeeping by ~85%.
  • Built a robust REST API integration using Django (backend) and JavaScript (frontend), enabling seamless synchronization of 10,000+ financial and inventory records with zero data loss.
  • Designed and implemented 5+ custom verification layers, achieving 99.8% data accuracy before syncing to KatanaMRP. Managed full project source code using Git and GitHub, maintaining clean version history and enabling reliable deployments to AWS EC2.
Teaching Assistant — Machine Learning — New York University, New York, NY Sep 2025 – Dec 2025
  • Guided 50+ graduate students through ML fundamentals, including preprocessing pipelining, supervised/unsupervised learning, Deep Learning, and model optimization techniques.
  • Conducted weekly office hours to debug Python code, explained algorithms, taught ML topics, and assisted with PyTorch implementations.
AI Engineer — Swaroop.ai, Ahmedabad, India Jan 2024 – Aug 2024
  • Enhanced TTS model performance by 25% through fine-tuning Coqui TTS across multiple Indian languages.
  • Improved STT accuracy by 18% via domain-specific audio datasets and custom preprocessing.
  • Integrated OpenAI Whisper into production, increasing transcription accuracy on noisy data by 30% and cutting inference latency by 20%.
Software Engineering Intern — Robotskull, Vadodara, India Oct 2023 – Dec 2024
  • Developed and maintained backend features for internal business applications, improving data processing workflows for sales and inventory operations.
  • Built RESTful API endpoints and integrated database queries to support inventory tracking, reporting, and operational analytics.

Projects

  1. ActualBudget Transaction Categorizer
    Python, FastAPI, PostgreSQL, MLflow, Docker, Kubernetes, Terraform.
    Built and deployed an ML-powered transaction categorization platform with data ingestion, experiment tracking, FastAPI serving, PostgreSQL storage, and GitOps-based infrastructure for reproducible model updates.
    [code]
  2. HexDrop (Secure File Transfer)
    Next.js, TypeScript, Prisma, PostgreSQL, AWS, Docker, K8s.
    This secure file-sharing application enables encrypted uploads and key-based downloads using AWS S3 and PostgreSQL.
    The platform operates on a full-stack DevOps pipeline featuring EKS orchestration, automated CI/CD, and scalable cloud infrastructure.
    [code]
  3. EyeConnect: Accessible Video Communication with AI-Powered Vision Assistance
    WebRTC, Supabase, OpenRouter AI, React, TypeScript.
    Accessibility platform connecting blind users with sighted volunteers via real-time video calls and AI vision assistance. Awarded 2nd place at NYU Hacks 2025.
    [code]
  4. Vision Transformer Optimization via Quantization & Efficient Attention
    PyTorch, bitsandbytes, FlashAttention-2, LoRA.
    Optimized ViT-L/16 using 4-bit/8-bit quantization and FlashAttention-2, achieving 4× model size reduction and 40% lower latency with minimal accuracy loss.
    [code]
  5. Model Merging for Large Language Models
    Python, PyTorch, Hugging Face, Google Colab.
    Implemented TIES and SLERP merging techniques to combine Mistral-7B variants with optimized hyperparameters for cross-task generalization.
    [code]
  6. Command Line Helper: Natural Language to Bash via Local LLM
    Converts natural-language instructions into bash commands using a local LLM and RAG-powered context retrieval, with both CLI and web interfaces.
    [code]
  7. Chapter: Secure Library Management System
    Python (Django), Oracle Data Modeler, HTML/CSS/JS, Oracle DB.
    Library management web application with role-based authentication, SQL-injection protection, and an employee dashboard for business metrics.
    [code]

A full list of repositories is available on GitHub.

Selected Certifications