Koushal Kishore Boju

Developed production AI microservices using Python (FastAPI) and C#/C++ on AWS to support LLM-powered workflows, reducing response latency by 31%. Built distributed AI workflows with LangGraph/LangChain, MCP, function/tool calling, RAG, semantic search, and vector stores (Pinecone, pgvector) to improve accuracy by 27%. Reduced token usage by 24% via prompt/context engineering, semantic caching, re-ranking, and model evaluation. Implemented scalable API pipelines integrating SageMaker, S3, EKS, Snowflake, Databricks, Kafka, and Airflow. Practiced production engineering with Docker, Kubernetes, Terraform, GitHub Actions, MLflow, and OpenTelemetry/Prometheus/Grafana, improving release time by 38%. Applied AI security and governance including guardrails, prompt-injection prevention, PII protection, and audit logging with human-in-the-loop validation.

Delivered backend AI services using Python with FastAPI/Flask and LangChain with Hugging Face, improving response accuracy by 26%. Built RAG and knowledge-retrieval services using Azure OpenAI embeddings and vector/graph retrieval tools (Qdrant, ChromaDB, FAISS, Weaviate) with hybrid search, reducing irrelevant responses by 32%. Implemented multi-agent enterprise workflows with LangGraph, LlamaIndex, CrewAI, Semantic Kernel, and tool calling. Improved latency by 29% using GPT-4/Claude/Gemini, prompt engineering, Redis caching, re-ranking, and LangSmith. Automated delivery using Docker, Kubernetes, Azure, Terraform, Jenkins, and GitLab CI/CD, reducing release turnaround by 35%. Added observability and production controls with OpenTelemetry, model evaluation, and AI guardrails (including prompt-injection prevention).

Built high-throughput backend microservices using Java (Spring Boot), Go, and REST/GraphQL/WebSocket APIs for multi-tenant SaaS, increasing throughput by 28%. Architected asynchronous event-driven services with Kafka, Redis, PostgreSQL/MySQL, MongoDB, Elasticsearch, and Oracle, reducing processing time by 31%. Optimized database access, caching, and distributed processing while integrating AWS (EC2, Lambda, S3, EKS) and Terraform, lowering API latency by 26%. Implemented tenant-aware APIs and reliability features such as rate limiting and reusable design patterns. Established CI/CD automation with Docker, Kubernetes, Jenkins/GitHub Actions, Git, and testing (Pytest/unit/integration), reducing deployment time by 37%. Monitored production using Prometheus, Grafana, Datadog, and OpenTelemetry and supported cross-functional delivery with product/QA/SRE/front-end teams.