Built production LLM and RAG systems using LangChain, LangGraph, AutoGen, enabling multi-agent reasoning and automation

Designed FastAPI-based AI services for real-time inference, deployed on Docker & Kubernetes

Implemented vector search with FAISS and Pinecone for low-latency, grounded LLM responses

Integrated AWS Bedrock and Vertex AI for model orchestration, routing, and inference

Applied MLOps / GenAIOps practices: monitoring, versioning, CI/CD, and model lifecycle management

Developed LLM-powered NLP solutions using BERT, RoBERTa, T5, and RAG pipelines

Built multi-step agent workflows to automate complex reasoning and retrieval tasks

Deployed AI services using FastAPI, Docker, Kubernetes, and cloud platforms

Focused on prompt engineering, evaluation, and monitoring for production reliability

Designed ML pipelines and backend APIs for analytics and AI use cases

Worked with classical ML + deep learning models and data platforms (Snowflake, Redshift)

Implemented CI/CD and MLOps workflows for model deployment and monitoring

Deployed ML and deep learning models using TensorFlow and PyTorch for predictive and real-time inference use cases

Built automated training pipelines with Airflow and MLflow and performed scalable feature engineering using SQL, Pandas, and Spark

Containerized ML services with Docker and deployed on Kubernetes with autoscaling and CI/CD (GitLab CI, Jenkins)

Implemented real-time inference pipelines using Kafka and Spark Streaming, with monitoring via Prometheus and Grafana