Varun Anusheel

AI/ML Engineer specializing in Generative AI, Computer Vision, and deploying advanced LLM solutions

Contact Me

About Me

Passionate about creating innovative AI solutions. Graduating with a Master’s in Artificial Intelligence from Northeastern University, I’ve led impactful projects like chatbot development at GE and fine-tuned language models at XNode. Recognized in GE’s guild competition, I thrive on transforming ideas into cutting-edge solutions in AI and Computer Vision.

01+ Years
experience
30+ Completed
projects
04+ companies
worked

Skills

I bring a diverse skill set in AI, software development, and cloud technologies, combining expertise in languages like Python, Java, C++, and R with powerful frameworks like PyTorch, TensorFlow, and Keras. Specializing in machine learning, deep learning, computer vision, and NLP, I create intelligent solutions using tools like Hugging Face, OpenCV, and Spacy. With experience in cloud platforms such as AWS and Google Cloud, I deploy scalable AI models using Docker, Kubernetes, and CI/CD pipelines. From building generative AI applications to developing robust APIs and deploying on cloud infrastructures, I focus on delivering innovative, production-ready systems that leverage the latest in AI technology.

Experience

Work
Education

Manipal Institute Of Technology

Computer & Communication
2017 - 2021

Northeastern University

Masters in Artificial Intelligence
2023 - Present

AI Intern

Krify Software Technologies Pvt Ltd
2017 - 2018

Full Stack Developer

EduMates - UK
2020 - 2020

Lean Performance Management Team

General Electric - USA
2021 - 2021

DevOps Engineer

General Electric - India
2021 - 2022

AI Engineering Intern

XNode
May 2024 - Dec 2024

GenAI/ML Software Engineer

Mythics LLC - USA
Jan 2025 - Aug 2025

Projects

LLM Wardrobe App

Full-stack wardrobe manager in React 19/TS which is LLM-powered outfit suggestions via Multimodal RAG, sub-3s delivery.

Github

MLOps for Supply Chain Automation

LSTM demand forecasting on VertexAI: 80% faster deployment via CI/CD, Airflow orchestration, auto-retraining.

Github

Deep Learning Image Restoration with PyTorch

CNN colorization using Lab color space: 73% accuracy, MSE 24, SSIM 0.78 for grayscale-to-color conversion.

Github

Fine-Tuning and Deploying LLaMA 3

QLoRA fine-tuned LLaMA 3 (7B→4GB): 76% sentiment accuracy, 4-5s inference on GCP Compute Engine.

Github

Generative Adversarial Neural Network

FashionGAN generates synthetic fashion images using a GAN trained on Fashion MNIST.

Github

Construction Project Management

Flutter app for 50+ employees: real-time expense/inventory tracking, 8% cost reduction, Firebase-backed.

Github

Real-Time Intruder Detection System

Raspberry Pi + YOLO security: 95% accuracy, 60% fewer false alarms, 15 FPS processing, <2s alert timeRetry.

Github

Fine-Tuning Vision Transformer (ViT)

ViT-B/16 fine-tuning: 92% accuracy on 5K images, 7% better than ResNet-50, 240min vs 360min training.

Github

LLM Evaluation with Guardrails

Automated LLM-as-judge with NeMo Guardrails and achieved 0.85 Pearson correlation, 45% fewer false positives, 94% reliability across 10K+ evaluations.

Github

Multi Modal RAG

Multimodal search using CLIP embeddings and Ollama for cross-modal image-text retrieval across 1M+ entries.

Github

LLM-Driven Medical Consultation

LangChain + MCTS chatbot with FAISS retrieval: 30% faster consultations, 40% better medical literature search

Github

ONNX Model Conversion

Converts Keras models (ResNet50, CIFAR-10) to ONNX, compares predictions, and visualizes training performance.

Github

Sentiment Analysis

FastAPI-based sentiment analysis app using Hugging Face model, classifying text as positive/negative with confidence score.

Github

Object Detection, Segmentation & Tracking

Mask R-CNN pipeline with SORT tracker: 87% accuracy, real-time multi-object tracking across 10K+ frames.

Github

ML Model with MLlib in pySpark

Built scalable ML pipeline with PySpark MLlib that reduced dimensionality 30% and cut execution time 35% through distributed data processing.

Github

Augmented Reality Product Visualization

2nd place GE competition: Unity AR app visualizing 100+ turbine components, 90s vs 6min manual reading.

Github

Breast Cancer Detection

Built a breast cancer classifier using ML algorithms (decision trees, logistic regression, SVM) achieving 90% accuracy on tumor detection.

Github

Event Booking Web Application

Flask/MySQL booking system for MAHE: 30% faster approvals, RESTful APIs supporting 5000+ concurrent users

Github

Recommender Systems

This repository implements collaborative, content-based, and demographic recommender systems for personalized user suggestions.

Github

MLOps Pipeline-CICD

Automates ML app deployment with MLOps, TensorFlow, Docker, and Kubernetes for model development and orchestration.

Github

Content-Based Image Retrieval (CBIR) System

Data analitical dashboard adaptable to all devices, with ui components and animated interactions.

Github

Movie Chatbot

Built an NLP-based movie chatbot using Python and Streamlit that allows users to ask questions and the chatbot will respond with relevant information.

Github

CLIP Image-Text Similarity

Using CLIP model to compute image-text similarity, comparing images with predefined text labels and searching for images matching a text query.

Github

Music Generator

Music Generation Web App creates both lyrics and music from user-provided prompts using GPT-NEO and Bark models.

Github

Grocery Store

Developed an full stack grocery store app using Android Studio and Firebase for real-time data management and seamless user experience.

Github

Image Viewer

Java image processor with GUI and CLI supporting 20+ operations (filters, transformations, compression) using MVC architecture and extensible I/O.

Github

Personal Portfolio Websiter

Designed a responsive portfolio website with HTML/CSS/JavaScript featuring interactive animations that increased inquiries by 15%.

Github

Have a new project idea?

We can connect to improve and develop that idea to a working project.

Contact me

Contact me

Get in touch

Call me

(+91) 9346075050
(857) 701-9025

E-mail

svarunanusheel@gmail.com
vikramsha.v@northeastern.edu

Location

Boston, MA