
Zaid A
Data Scientist
Kompetenzen

Meine Dienstleistungen

Portfolio
Arbeitserfahrung
Senior Data Scientist
Microsoft • Vollzeit
Feb 2026 - Present • 7 mos
Architected Real-Time Multimodal Agents: (real-time video and voice AI agents) leveraging WebRTC/Daily infrastructure, implementing real-time barge-in handling and interruption detection for natural user interaction. Latency & Performance Optimization: Slashed end-to-end voice/video agent latency by 68% (from 14 seconds down to 4.5 seconds), enabling conversational real-time response times. Vector Database Architecture & Cost Reduction: Migrated search infrastructure from Azure AI Search to Weaviate, achieving a 10x reduction in infrastructure costs while preserving retrieval accuracy and search throughput. Frontier LLM Benchmarking & Evaluation: Performed frontier-level evaluations comparing standard LLMs against specialized agentic frameworks (Daily / Command-driven systems), executing trajectory auditing, tool-call accuracy checks, and hallucination evaluations. Marketing Attribution & Lift Modeling: Conducted multi-touch marketing attribution and incrementality lift studies to optimize vendor ad spend allocation, identifying highest-ROI acquisition channels for enterprise clients.
Advanced Data Science Associate
ZS • Vollzeit
Apr 2024 - Present • 2 yrs 5 mos
Serving as an Advanced Data Science Associate in Bengaluru, India. Responsible for leveraging advanced analytics and data science methodologies to solve complex business problems. The role involves applying expertise in generative AI and machine learning within a professional services environment.
data scientist
Intellect IT • Vollzeit
Jan 2023 - Apr 2024 • 1 yr 3 mos
Built an Intelligent Document Processing (IDP) pipeline leveraging LayoutLM and fine-tuned BERT models, reducing end-to-end processing times from 4 hours to under 2 minutes. Awarded the Mr. Wiki Award for Versatility for cross-functional agility across complex computer vision, NLP, and backend infrastructure layers. Developed a customer churn prediction model with 88% accuracy, cutting quarterly churn by 20% and preserving $15M in Assets Under Management (AUM).