Sayed Sohail Peerzade

I am an AI Engineer and Software Engineer with experience building production-ready AI applications, real-time systems, and full-stack platforms. My work sits at the intersection of LLM engineering, backend architecture, frontend product development, and applied machine learning.

I am currently working as a Founding AI Engineer at XRaceAI, where I am building a real-time AI race coaching platform that processes live telemetry from racing simulators and delivers low-latency, in-session driver feedback. I have worked on telemetry ingestion pipelines using FastAPI, WebSockets, and Kafka; Graph-RAG systems using Neo4j; real-time voice coaching with LiveKit, STT/TTS, WebRTC, and LLM orchestration; and AI cost/latency optimisation through task-based model routing, caching, and token optimisation.

Previously, I worked as an AI Engineer Intern at Insight Dev, where I helped build a Legal AI platform using GPT-4, Cohere, LangChain, RAG, document chunking, semantic search, and AI-powered summarisation. I also evaluated RAG and summarisation systems using metrics such as ROUGE, BLEU, ARES, and RAGAs. Alongside this, I have experience as a Data Science Consultant at Outlier, working on RLHF, AI model evaluation, ranking frameworks, Python/SQL code review, and improving AI-generated outputs.

My technical background includes Python, TypeScript, JavaScript, React, Next.js, Angular, Flask, FastAPI, SQL databases, MongoDB, AWS, Docker, Kubernetes, Jenkins, GitHub Actions, and testing tools such as Cypress, Jest, PyTest, Playwright, and Selenium. In AI and machine learning, I have hands-on experience with LLMs, RAG, Graph-RAG, AI agents, prompt engineering, embeddings, vector search, RLHF, classical ML models, clustering, classification, neural networks, and model evaluation.

I am especially interested in roles involving AI engineering, LLM applications, RAG systems, AI agents, real-time AI products, backend engineering, and full-stack product development. I enjoy building systems that are practical, scalable, measurable, and genuinely useful to users.