Software engineer, technical writer & open-source maintainer
I’m Avdhesh Varshney, an experienced frontend and machine learning engineer passionate about crafting intelligent, user-friendly applications—merging sleek, responsive interfaces with robust AI solutions to create impactful tools for developers and the world.
Contrbution Graph
Work Experience
QuickSell & DoubleTick
Software Developer (SDE I)
Built enterprise authentication features—SSO (SAML/OIDC + Google Workspace), improved onboarding, and CSP/COOP compliance. Delivered major bot-builder upgrades like Circular Linking, LeadSquared integration, auth-key webhooks, and fixed 15+ workflow issues. Optimized media with tokenized URLs and scroll-based unloading, cutting chat memory by 60%. Closed key security gaps and shipped AI messaging tools improving response speed by 25–35%.
Girlscript Foundation
Project Admin & Mentor
Led and mentored 100+ contributors across open-source projects collaborative blogging platform (React, Node, MongoDB) and ai automation systems (ML, DL, LLM). Designed contribution roadmaps and reviewed PRs to implement new UI components, API routes, and system optimizations. Guided students on complex technical tasks: integrating Fine-tuned LLMs & RAG pipelines, wrapping CNN/GAN models into Flask/Node inference APIs, and improving model preprocessing and architecture for production-ready deployment.
Winter of Code
Machine Learning Developer
Delivered 3 major CV pipelines — Gender Classification, Image Deblurring, and Cartoonification using TensorFlow, Keras, OpenCV, Flask, and Streamlit. Built end-to-end apps with custom preprocessing (32×32/64×64 transforms), model loading (H5/JSON weights), CNN/autoencoder inference, and UI workflows with downloads and live previews. Improved model stability, accuracy, and UX across repositories and secured 1st rank for consistently production-ready ML contributions.
Social Winter of Code
Project Mentor
Mentored 25+ contributors across AI/ML repositories including Face-X, DL-Simplified, ML-Crate, ML-ProjectKart, GeetaGPT, and AI-Code, reviewing 80+ pull requests and driving measurable project progress. Guided contributors through building and refining chatbot conversational pipelines and RAG-based data integration with the Bhagavad Gita text, improved deep-learning workflows (SMOTE preprocessing, dropout regularization, Adam optimization, accuracy evaluation) for ML models, ensured robust testing and refined repo architecture for production-ready AI systems.
Code Peak
Open Source Developer
Delivered 30+ contributions across ML, automation, and full-stack repos. Built predictive pipelines (nurse-stress, bug-analysis notebooks) using Python, NumPy, Pandas, Plotly, and Matplotlib. Shipped automation scripts using Instabot, Telebot, and GitHub python packages. Implemented Firebase Google Sign-In, Resend-based email subscriptions, and Cloudinary uploads. Refactored frontend/backend architecture, improving modularity, build efficiency, and repo consistency.
Hacktoberfest
Open Source Developer
Contributed ML and deep-learning models across repos like ML-Crate, DL-Simplified, and ML-ProjectKart, shipping end-to-end pipelines for image classification, voice-based gender recognition, celebrity face detection, cryptocurrency analysis, and multiple health-risk prediction datasets. Performed extensive EDA, feature engineering, correlation mapping, model benchmarking (LR, Ridge, Lasso, DNNs), and built DL architectures (InceptionV3, custom CNNs) with data augmentation and visualization workflows.
Girlscript Summer of Code
Open Source Developer
Secured 6th rank among 1500+ active contributors and 17,000+ applicants. Contributed across multiple repos like GitHub-Automation-Scripts, Amazing-Python-Scripts, Pentesting-and-Hacking-Scripts, GameZone, and Dev-Geeks, shipping 70+ merged PRs. Built automation scripts, mini-apps, UI components, and logic-based games using Python, JavaScript, and Web APIs. This experience strengthened my engineering fundamentals and pushed me toward building scalable automation workflows and ML/AI-driven tools.
Social(Formerly Script Foundation)
Open Source Developer
Secured 1st rank. Contributed 45+ merged PRs across distributed repos with consistent CI-clean code. Built 20+ functional web apps (ChatGPT Clone, Video Player, Piano, Meal Generator, etc.) using HTML/CSS/JS and optimized DOM, async ops, and event pipelines. Implemented graph and number-theory algorithms (Fleury, Hierholzer, Lucas-Lehmer, D-Heap) in multiple languages. Closed 80+ tracked issues, improved repository structure, fixed edge cases, and delivered production-ready modules with high acceptance rate.