lakshit sachdeva

building stuff at the intersection of machine learning and scalable systems. currently working on things that learn, reason, and optimize themselves.

some things i've built

drift-ml: LLM AutoML Agent

LLM-guided AutoML agent that uploads CSV files, chats with AI, and gets trained models. Features beautiful charts, model comparison, smart predictions, and Gemini-powered pipeline generation.

PythonTypeScriptAutoML

Vision-AI Waste Management

Deployed a custom ResNet-50 backbone with 95.45% inference accuracy. Engineered real-time edge processing on Arduino for automated industrial waste classification.

TensorFlowOpenCVEdge AIIoT

Predictive Alpha: Quant Platform

Developing a full-stack financial engine with real-time volatility tracking, secure multi-tenant architecture, and automated portfolio optimization algorithms.

Next.jsSupabaseSQLQuantitative ML

QuizGen - AI-Powered Quiz App

Android quiz application with AI-powered question generation from notes, PDFs, and images. Features preloaded quizzes, Gemini API integration, Firebase authentication, and ML Kit OCR.

AndroidJavaFirebaseGemini APIML Kit

CAPTCHA-X Recognition System

Deep learning system using convolutional neural networks for CAPTCHA recognition. 24-layer CNN architecture achieving 98.97% accuracy on 5-character CAPTCHAs with grayscale preprocessing.

PythonTensorFlowCNNComputer Vision

Claims Processor

Intelligent insurance claims processing system leveraging Cloud LLM APIs, RAG, and Gmail integration. Automated claim evaluation with structured query parsing, confidence scoring, and FAISS-powered semantic search.

PythonFastAPIRAGFAISSStreamlitGmail API

wanna build something?
let's connect

say hi