Saksham Poply

I am a generative AI-focused software engineer specializing in building scalable web applications and retrieval-augmented search systems.

At EPAM, I developed 3+ generative AI projects using LangChain and LangGraph, integrating OpenAI, DeepSeek, Claude, and LLaMA to improve response accuracy by 30%. I also implemented model fine-tuning and prompt engineering that reduced hallucinations by 25% and boosted LLM response efficiency by 35%.

As a research intern at ISRO Space Applications Centre I designed a lunar visualization system with Blender and Unreal Engine 5, automated the generation and storage of 1000+ OBJ files, and produced 500+ 3D simulated images for scientific analysis, reducing manual effort by 40%.

I build end-to-end AI products—deploying Flask, FastAPI, and Django backends, optimizing API response times by 40%, publishing RAG research that improved media search relevance by over 40%, and shipping user-facing interfaces that increased engagement and reduced bounce rates.