RISHAB PAL

Enhanced eCommerce site search relevance and ranking metrics, boosting click-through rates (CTR), conversion rates, and ad revenue. Innovated search keyword generation with Milvus-based vector search. Optimized retail product pricing by employing ML demand prediction models. Pioneered customer churn prediction for Makro retail, utilizing diverse sales data to design KPIs and develop predictive models.

Revamped measurement automation for US residential houses by engineering a high performance Line Detection model. Developed segmentation models for Roof & Wall Facet detection. Improvised complex PyTorch models to facilitate ONNX & TensorRT export for efficient deployment using NVIDIA Triton Server. Introduced best MLOps practices to automate testing, deployment, and monitoring.

Created an innovative conversational assistant using deep learning to resolve queries for over 400 million monthly active users. Led the re-engineering of a Face-Dedupe KYC platform, including Face Recognition, Face Matching, and Face Liveness Check models. Actively involved in the interview process for candidates at various seniority levels.

Developed VirtualHR, a tool that automates matching job descriptions with resumes using FastText, NER, and LayoutLM. Deployed the world's first Privacy Enhanced Gateway for FortressIQ with a custom OCR solution. Prototyped an NLP based chatbot using RNNs to parse guest comments and assess sentiments.

Completely owned Public Personal Mobility - AI Conductor project. Implemented Kalman Filter based Multi-Object Tracking, Hand Gesture Recognition using ConvLSTM, and Single-Shot Face Recognition. Proposed and implemented predictive analysis for retail sales forecasting using ARIMA and Exponential Smoothing.