Jan 2021 – Present · 5 yrs 2 mos

• Led development of an Agentic AI application designed to automate complex research for enterprise use cases, driving a significant increase in analytical efficiency.
• Directed R&D and implementation of advanced observability platform to enhance the tracking, monitoring, and management capabilities for production Generative AI, AI/ML, and Deep Learning (DL) applications, utilizing MLflow and Prometheus.
• Engineered and implemented performance optimization of AI applications and models using a mix of hardware-agnostic and hardware-specific technologies, including Ray Serve, OpenVINO and PEFT, to minimize inference latency and maximize system throughput.
• Oversaw the integration of Large Language Models (LLMs) using commercial services such as Azure OpenAI and Amazon Bedrock, orchestrating conversational agents with Autogen and serving models via high-performance APIs like FastAPI and Uvicorn.
• Built, mentored, and led a high-performing team of young technical professionals, successfully fostering their technical expertise and ensuring the delivery of mission-critical AI/ML solutions.

Dec 2012 – Dec 2021 · 9 yrs

• Collaborated directly with customers and domain experts to develop and optimize High-Performance Computing (HPC) applications on advanced CPU and GPU platforms.
• Specialized in deep code optimization for high-end hardware, including GPUs and Xeon/Xeon-Phi processors, utilizing comprehensive parallel programming technologies such as CUDA, OpenMP, MPI, HIP, SYC and OneAPI.
• Enhanced AI/ML training and inference performance on multi-core CPUs and GPUs through implementing manual parallelism techniques and leveraging specialized tools like Horovod and OpenVINO.