Open to opportunities

Hemanth Kumar

@hemanthkumar11

Message

I build production LLM, RAG, and AI-agent systems that improve verification, research, and compliance workflows.

India

Message

What I'm looking for

I'm looking to build production AI systems involving LLMs, RAG, AI agents, and backend services, where I can improve reliability, cost efficiency, and real-world user outcomes.

At Novetum, I build and scale a FastAPI AI verification API powered by Gemini, serving 3,000+ requests per day while reducing inference costs by 35% through model selection and batching.

I've designed agentic research pipelines with function-calling, web search, MCP tools, self-review loops, and structured outputs. I also build RAG products such as Ansvara, AppForge, and ARIA using LangChain, vector databases, FastAPI, PostgreSQL, and observability tooling.

Previously at Celebal Technologies, I built an unsupervised NLP pipeline using TF-IDF, K-Means, and LDA to extract structured insights from large-scale text. I contribute to open source by improving LangChain provider streaming resilience and checkpoint recovery for hyperparameter optimization.

Experience

Work history, roles, and key accomplishments

Current

AI Engineering Intern

Current

Novetum

Jun 2026 - Present (2 months)

Built and scaled a FastAPI AI verification API powered by Gemini, serving 3,000+ requests/day and cutting inference costs by 35%. Designed an agentic LLM pipeline for autonomous research and owned prompt workflows with structured-output schemas.

fastAPI Gemini Langfuse Function Calling MCP Prompt Engineering A B Testing Python

Data Science Intern

Celebal Technologies

May 2025 - Jul 2025 (2 months)

Built an unsupervised NLP pipeline using TF-IDF, K-Means, and LDA to extract structured insights from large-scale text, achieving 0.79 homogeneity and 0.71 silhouette score.

Python NLP TF IDF K Means LDA scikit learn Pandas NumPy