Vishanth Iyer

About

ML/NLP engineer with experience at NVIDIA and Cruise building production-grade AI systems across genomics/biomedical research and autonomous vehicle data. Has delivered multimodal LLM pipelines, large-scale entity resolution, and hybrid semantic search (BERT embeddings + FAISS + Elasticsearch), with measurable impact (≈40% accuracy/retrieval gains; ≈30% data consistency improvement) and strong MLOps practices (Kubernetes, CI/CD, MLflow, Prometheus/Grafana).

Key Strengths

• Built multimodal LLM system linking text, images, and time-series biomedical/genomic data
• Designed custom preprocessing/validation pipelines to handle inconsistent data formats and quality (PySpark, TFDV)
• Delivered large-scale entity resolution using rule-based + ML (XGBoost) and embedding similarity; validated with gold dataset and precision/recall/F1
• Improved downstream performance via A/B testing after data cleaning; reduced redundancy and improved consistency ~30%
• Implemented semantic search at scale with fine-tuned BERT embeddings + FAISS + Elasticsearch hybrid retrieval; improved research retrieval speed ~40%
• Production-grade ML/data workflow engineering: orchestration (Airflow/Kubeflow), distributed compute (Spark/Dask), containerization (Docker/K8s), CI/CD (GitHub Actions), model/data versioning (MLflow), monitoring (Prometheus/Grafana)
• Strong prototype-to-production decision-making using accuracy, stability, latency targets, ROI, and operational reliability

Experience

Education

Skills

Languages