I am Pratik Sheth, a Senior AI Research Engineer with 7+ years of applied research experience in computer vision, machine learning operations, and distributed systems architecture. Currently serving as AI Engineer III at Polymerize’s Material Informatics Platform, where I lead the Image Analysis division and spearhead the development of production-grade AI systems.
My expertise bridges cutting-edge AI research with enterprise-scale deployment. I specialize in building production RAG pipelines that transform how researchers interact with scientific literature, designing computer vision systems that analyze thousands of microscopy images daily, and implementing LLMOps infrastructure using modern frameworks like LangChain, HuggingFace Transformers, and LoRA fine-tuning.
Research Areas & Technical Expertise:
Machine Learning Research & Development
- Computer Vision Research: Advanced architectures including ResNet, YOLO, SAM, and Diffusion Models for materials science applications
- Large Language Models: LLMOps pipeline development, fine-tuning methodologies, and retrieval-augmented generation systems
- Multimodal AI Systems: Integration of vision and language models for document intelligence and scientific literature analysis
- Neural Architecture Optimization: Model compression, quantization, and distributed training for production deployment
Applied Research Infrastructure
- Experimental Platforms: Cloud-native ML infrastructure design with focus on reproducibility and scalability
- High-Performance Computing: GPU cluster management and distributed computing frameworks for large-scale experimentation
- Data Processing Pipelines: Real-time streaming architectures for continuous learning and model adaptation
- Research Software Engineering: CI/CD systems for ML research with automated experimentation and results tracking
Technical Stack:
- AI/ML Frameworks: PyTorch, TensorFlow, HuggingFace Transformers, LangChain, LlamaIndex
- LLM & NLP Tools: OpenAI API, Anthropic Claude, Ollama, vLLM, LoRA, QLoRA, PEFT
- Computer Vision: OpenCV, YOLO, SAM, Diffusion Models, Detectron2, MMDetection
- Vector & Databases: Weaviate, FAISS, PostgreSQL, MongoDB
- MLOps & Deployment: MLflow, Weights & Biases, NVIDIA Triton, TensorRT, Docker, Kubernetes
- Cloud & Infrastructure: AWS (EKS, SageMaker, Lambda), Terraform, Helm Charts, FastAPI
I hold an MBA in Technology Management and a B.E. in Information Technology from Gujarat Technological University. My research interests include multimodal AI systems, neural architecture optimization, and scalable machine learning infrastructure. My work demonstrates expertise in translating theoretical advances into production-grade implementations across materials informatics, document intelligence, and real-time processing systems.
This portfolio showcases my journey through complex research challenges and innovative solutions in the rapidly evolving landscape of AI research engineering and production ML systems.
Connect with me at ps@outlook.in for collaboration opportunities, research discussions, or technical consulting.