Prabhat Kumar Gupta - Senior AI Engineer

Prabhat
Senior AI Engineer. LLM Systems Builder

Senior AI Engineer with 6.2+ years of experience building LLM-powered products, RAG systems, and agentic AI platforms from ground up. Currently driving hybrid search, knowledge orchestration, and VLM-based systems at AI71 (Abu Dhabi).

Get in touch

Download Resume

Work Experience

AI71 - RAG & Knowledge Orchestration

Designed and optimized Vector and Hybrid Search pipelines for large-scale document retrieval. Built scalable Knowledge Base supporting 7+ document types. Reduced API latency by 60% (5s to 2s).

Learn More →

Zepto - ML Model Serving Platform

Created extensible ML model serving platform using TF-Serving, reducing latency by 99% (400ms to 5ms). Built VLM-powered product filter generation for 10k+ products using GPT-4o.

Learn More →

UnifyApps - Banking LLM Chatbot

Fine-tuned Llama3-8B-Instruct for banking chatbot scenarios achieving 90%+ accuracy. Built Text-to-SQL AI Copilot for Salesforce DB, reducing query creation time by 95%.

Learn More →

Sprinklr - ML Infrastructure

Built Agent Scoring mechanism using distilBERT, driving 100% product revenue growth. Secured 4000+ ML deployments with IAM-based access control. Optimized Kafka pipeline reducing false alerts by 70%.

Learn More →

Recommendations & Kudos

Engineer at AI71

Senior Engineer, AI71

Prabhat is an exceptional AI engineer who joined as the 2nd AI engineer in the company. His contributions to core architecture design, ML system planning, and RAG pipeline integration have been invaluable. He consistently delivers scalable, production-ready solutions that create tangible business value.

Team Lead at Sprinklr

Engineering Manager, Sprinklr

Prabhat's work on the Agent Scoring mechanism using distilBERT was instrumental in driving 100% product revenue growth over 2 years. His expertise in MLOps, Kubernetes, and infrastructure security helped secure 4000+ ML deployments. He's a true leader who excels at rapid prototyping and delivering scalable solutions.

Associate Director at Zepto

ML Platform Lead, Zepto

Prabhat created a highly extensible ML model serving platform that reduced latency by 99%. His work on VLM-powered product filters revolutionized our search capabilities. He's known for his ability to build production-ready systems that scale efficiently.

Engineer

Software Engineer

Working with Prabhat has been an incredible learning experience. His expertise in LLMs, RAG systems, and agentic AI is unmatched. He's a great mentor who shares knowledge freely and always pushes for best practices in ML engineering and system design.

Technical Expertise

LLMs & RAG Systems

GPT-5, Claude-Sonnet-4, BAAI/BGE-m3, DeepSeek-OCR, LoRA, Mistral. Building production RAG pipelines and knowledge orchestration systems.

ML Frameworks

PyTorch, LangChain, LlamaIndex, TensorFlow, Keras, Transformers, Accelerate, DeepSpeed, PEFT.

MLOps & Infrastructure

VLLM, MLC, Docker, Kubernetes, Kafka, GitHub Actions. Building scalable ML infrastructure and deployment pipelines.

Cloud Platforms

AWS, Google Cloud, Azure. Deploying and managing ML workloads across multiple cloud providers.

Agentic AI

ReAct-based Agentic Orchestrator, multi-agent workflows, dynamic workflow management. Increasing use-case coverage by 5x.

Fine-tuning & Evaluation

RAG, fine-tuning, and benchmarking frameworks. Precision@k, Recall@k, answer-correctness, faithfulness metrics.

Get in touch

Projects & Articles ✨

🚀

Open-Cursor - Free AI Coding Assistant

An open-source version of Cursor coding agent that runs locally, fully powered with open-source LLMs. Privacy-first, cost-free, and fully customizable AI coding assistant.

GitHub Article

Building Production RAG Systems

Insights on designing and optimizing RAG pipelines for large-scale document retrieval and knowledge orchestration.

Read on Medium →

MLOps Best Practices

Lessons learned from deploying 4000+ ML models in production. Infrastructure security, scalability, and reliability.

Read on Medium →

Agentic AI & Multi-Agent Systems

Building ReAct-based agentic orchestrators and multi-agent workflows for dynamic use-case coverage.

Read on Medium →