About

👋 Hi, I’m Pooja

I’m a Machine Learning Engineer who gets excited about turning ambitious ML ideas into production systems that actually work at scale. I live at the intersection of data engineering, applied machine learning, and MLOps—basically, I’m happiest when I’m wrangling distributed systems and making models perform in the wild, not just in notebooks.

🛠️What I Build

I design end-to-end ML architectures that handle real traffic. Recently, I’ve been deep in RAG systems, integrating OpenAI GPT-4, Pinecone, Cohere Reranker and Snowflake to power semantic search and recommendations that actually understand context. I’ve fine-tuned diffusion models with LoRA, optimized multimodal embeddings using CLIP, and built FastAPI inference microservices that scale horizontally with Docker and Kubernetes.

Before going all-in on AI systems, I spent years building rock-solid data infrastructure—think Spark, Hive, and AWS EMR processing terabytes of data, with MLflow and Airflow keeping everything reproducible and automated. I genuinely enjoy the full stack: from messy data ingestion and feature engineering, through model training and experimentation, all the way to deployment and monitoring.

💡What Drives Me

I’m fascinated by retrieval systems, distributed training, vector databases and making generative models actually useful (not just impressive). My goal isn’t just to build models that work in theory, it’s to create high-availability AI platforms that survive production, scale gracefully, and deliver measurable business value. I believe the best ML systems are built by people who understand both the algorithms and the infrastructure. That’s where I thrive.

🎓Certification

AWS Certified Machine Learning – Associate | Experienced in architecting ML solutions on AWS infrastructure

🔧Tech I Work With

ML & AI: PyTorch, TensorFlow, Transformers, Diffusion Models, CLIP, LoRA, RAG MLOps: MLflow, Kubeflow, Airflow, Docker, Kubernetes, CI/CD Data: Spark, Hive, Snowflake, AWS EMR, Kafka Vector/Semantic Search: Pinecone, FAISS, Cohere Cloud & APIs: AWS, FastAPI, REST, microservices

Always open to collaborating on challenging ML systems problems, especially if they involve making things work at scale. 🚀

Feel free to explore my projects and blog posts, or connect with me on LinkedIn and GitHub!