KARTHIK GOVINDAPPA

Directed and optimized operations for a chemical distribution business, overseeing a team of 10 across sales, inventory, and logistics to ensure efficient supply chain management.

Spearheaded the development and deployment of an LLM-based SDS compliance agent, automating critical regulatory checks and enhancing operational efficiency.

Developed a commodity price forecasting system utilizing chemical market data to inform strategic pricing decisions and optimize revenue generation.

Engineered and deployed ranking and personalization models for Mobee, boosting NDCG@10 by 15% (offline) and CTR by 10% (production A/B tests).

Built the core candidate generation stage for product matching, processing millions of products using transformer embeddings and ANN retrieval to generate high-recall sets.

Designed and implemented a company-wide A/B testing and experiment-assignment platform, enabling data-driven evaluation of model variants in production.

Developed a dynamic pricing model that drove a 12% revenue increase and 8% higher conversion through strategic integration with raw event streams.

Productionized ML models across the team using FastAPI, Kubernetes/Helm, and AWS (Terraform) for scalable and repeatable deployments.

Independently developed and deployed a suite of four genomics pipeline packages (variant calling, ChIP-seq, RNA-seq, single-cell), adopted as core infrastructure by a cancer-research lab.

Designed config-driven CLIs to download controlled-access NIH data (dbGaP, GDC) and orchestrate multi-stage workflows on a high-performance computing cluster (SLURM).

Architected each pipeline stage as an independently-resourced batch job, ensuring reproducible, large-scale genomic processing on Rutgers' Amarel cluster.

Designed and deployed a driver drowsiness detection and glance classification system, achieving 15 FPS on edge devices with an improved face detection model.

Led research into optimizing ML models for low-power edge hardware, including Intel Movidius, Snapdragon SoC, and Nvidia Jetson, to enhance real-time performance.

Engineered an ensemble model (neural networks, SVM, Random Forest) for robust driver behavioral state detection from vehicle telemetry data.

Built an air quality forecasting system using sensor data from an indoor home robot, providing real-time environmental insights.