Praveen Kumar Balaji

About Praveen Kumar Balaji

Hi, I’m Praveen Kumar Balaji, an AI/ML Engineer with 5+ years of experience building and deploying production-grade machine learning systems across NLP, LLMs, and scalable data platforms. I focus on turning complex data and modeling challenges into reliable AI solutions that deliver measurable business impact.I’ve worked across startup, consulting, and enterprise environments, owning the end-to-end ML lifecycle—from data engineering and model development to deployment, monitoring, and optimization. My recent work includes RAG pipelines, transformer-based NLP systems, prompt-engineered LLM workflows, and cloud-scale MLOps, improving model accuracy by up to 32% and reducing inference costs by 52%.My core skills include Python, SQL, PySpark, TensorFlow/PyTorch, Hugging Face, Docker, Kubernetes, and AWS/Azure/GCP. I enjoy collaborating with cross-functional teams and communicating clearly with both technical and business stakeholders. Available for onsite, hybrid, or remote roles and open to relocation within the U.S.

Experience

1. Founding Ai Ml Engineer(Generative Ai)

   Whyhow.Ai

   Jun 2025 — Present · CA, US

   Built and enhanced LLM/agent orchestration (Planner/supervisor patterns, tool-using agents, routing, guardrails).*Implemented intent classification information extraction validation and decision logic for servicing workflows• Architected enterprise-scale AI platform processing 12M+ FDA medical device adverse event records using Snowflake data warehouse, implementingdistributed LLM pipeline with GPT-4, Claude, and Hugging Face Transformers achieving 94% accuracy in automated safety classification while reducinginfrastructure costs by 70% through optimized model deployment and efficient batch processing strategies.• Developed production-grade NLP system using Hugging Face Transformers and custom BERT/BioBERT models for multi-agent AI architecture,processing unstructured medical narratives with fine-tuned RoBERTa and DistilBERT for medical entity recognition, achieving 91% precision in productliability case assessment and reducing attorney review time by 80% through automated evidence extraction.• Built production RAG system with Pinecone vector database processing and semantic embeddings using sentence-transformers and customembedding models, implementing real-time adverse event pattern matching with FAISS similarity search and LangChain orchestration, enabling sub-second query response across terabyte-scale healthcare datasets for regulatory compliance automation.• Deployed enterprise-grade MLOps pipeline using Docker containers and Kubernetes orchestration for scalable AI model serving, implementingcontinuous integration with Hugging Face Model Hub, MLflow experiment tracking, and automated A/B testing frameworks, maintaining 99.9% uptime

Education

• George Mason University

  Master's degree, data analytics

• Sathyabama Institute of Science & Technology, Chennai

  Bachelor of Engineering - BE, Electrical, Electronics and Communications Engineering

  2016 — 2020