Saneha Siddique

I’m an AI/ML and GenAI engineer with hands-on experience in CV, genAI, and agentic AI, plus full-stack development. I’m driven by turning complex model capabilities into scalable, reliable applications that solve real problems for real users.

In my Machine Learning Engineer internship at Scraperrs Lab, I built YOLO-based product detection pipelines (preprocessing, augmentation, annotation) achieving 90%+ mAP and deployed the solution via FastAPI. I also developed iterative agent nodes in LangGraph for a WhatsApp bot—integrating Weaviate for efficient data retrieval and context management—while training multiple branded object detection models across 50+ product categories.

I’m equally focused on engineering quality and maintainability. At 10Pearls, I built a scalable internal application using React.js, Node.js/Express, and MongoDB, and achieved 100% test coverage with Jest (frontend) and Mocha/Chai (backend). I improved code maintainability with Pino Logger and configured SonarQube to reach over 90% code quality.

My projects reflect how I like to work: multimodal + agentic + user-oriented. For AsaanForm, I built a web-based multi-agent system for bilingual (English/Urdu) OCR form extraction and semantic auto-fill with validation and missing-field resolution via a RAG chatbot. I’ve also developed QuranAI (LLM fine-tuning + RAG with FAISS, JWT sessions, OTP verification) and a Digital Prescription & Medication Reminder app with secure role-based dashboards and QR-based prescription verification.