Looking for a job

Deepak Kumar

@deepakkumar94

Senior Software Engineer at Clearwater Security building an async LLM orchestration pipeline with FastAPI, LangGraph, and AWS Bedrock.

India

Message

At Clearwater Security, I built an async LLM orchestration pipeline using FastAPI, LangGraph, and AWS Bedrock, with parallel execution, checkpointing, and retry handling.

At Krista AI, I re-architected RAG retrieval with BM25 and RRF, improving file ranks by 25.6% and chunk ranks by 44% across 1,500 production benchmark queries. I also designed model versioning for 400 ML models and caching and async inference architecture that improved response time by 40% under peak production load.

Experience

Work history, roles, and key accomplishments

Current

Senior Software Engineer

Current

Clearwater Security

Apr 2026 - Present (7 months)

Built an async LLM orchestration pipeline using FastAPI, LangGraph, and AWS Bedrock with parallel execution, checkpointing, and retry handling.

fastAPI LangGraph AWS LLM Async Programming

Data Scientist

Krista AI

Nov 2024 - Mar 2026 (1 year 4 months)

Re-architected RAG's retrieval architecture using BM25 + RRF, improving file ranks by 25.6% and chunk ranks by 44%. Designed robust model versioning system for 400 ML models and scalable caching and async inference architecture, improving end-to-end response time by 40% under peak load.

RAG BM25 RRF Model Versioning Caching Celery Async Inference Fcntl

Software Engineer

Krista AI

Jul 2022 - Oct 2024 (2 years 3 months)

Built no-code ML training infrastructure for FastText text classification models. Implemented model migration system for scikit-learn upgrades across 460+ models, fine-tuned GPT-4o Mini for structured JSON output, and built RAG system using BLOOM-7B and ChromaDB.

FastText scikit learn GPT 4o Mini RAG BLOOM 7B ChromaDB Java Testing