Asfar Rizvi

Led architecture of enterprise GenAI/LLM systems, building RAG-based assistants that automated 68% of Tier-1 support queries. Implemented sub-second inference APIs, 100M+ event feature pipelines, and MLOps workflows (MLflow, drift detection, retraining), improving reliability by ~18%.

Built a geospatial ETA platform predicting truck wait times with 20-minute accuracy, cutting logistics dwell times by 15% and productizing real-time Facility Insights used for dispatch optimization.

Built enterprise conversational AI assistants using RAG to automate 65–75% of HR and support queries, reducing response time from 12s to 4.8s and lowering operational workload by ~30%. Scaled semantic retrieval with embeddings/FAISS (300ms to 90ms) and implemented hallucination detection and response validation evaluations.

Deployed a multilingual LLM-based virtual assistant resolving ~70% of inquiries and reducing response times by ~70% via a RAG framework and multi-agent orchestration for production conversational AI.

Delivered ML/NLP and recommendation solutions, planning and deploying scalable ML pipelines processing 10M+ records daily and reducing model deployment time by ~40%. Built real-time and batch inference systems, improving reliability and throughput through feature engineering and experimentation aligned to ML KPIs.

Produced NLP pipelines for intent detection, semantic search, and knowledge retrieval, building conversational AI for enterprise IT support. Achieved ~70% ticket auto-resolution and reduced support handling time by ~50% using intelligent workflow routing and scalable backend systems.

Re-architected a real-time recommendation pipeline increasing CTR by 60% and revenue per visitor by 10% through embeddings, Google Recommendations AI prototyping, and automated retraining pipelines.

Co-built ML components for Uber Eats recommendations and support automation, improving home-feed CTR and reducing support handling time via GBDT and early deep-learning models.

Built recommendation and ranking systems that increased booking conversion by 9% using personalized ranking models. Improved CTR by up to 60% with collaborative filtering and embedding-based approaches, and increased revenue per visitor by ~10% through personalization, A/B testing, and large-scale Spark data pipelines.

Built an early Location Relevance search model that improved booking conversions and reduced irrelevant results by engineering geo and listing features and deploying reproducible data pipelines.

Completed an MS in Computer Science coursework including statistical inference, machine learning, time-series forecasting, and data analysis with Python.

Completed a Master of Science in Computer Science focusing on advanced machine learning and data-driven systems.

Earned a BS in Computer Science (Data Science) with coursework in data structures and algorithms, database systems, artificial intelligence, and distributed systems.

Earned a Bachelor of Science in Computer Science with a concentration in Data Science focusing on applied machine learning and data engineering.