sarthak arora

I’m a senior software engineer with 4+ years building cloud-native platforms, full-stack apps, and production AI across healthcare and SaaS, with deep AWS experience. I’m now focused on agentic AI and RAG pipelines that hold up in real enterprise workflows.

As Co-Founder & CTO at AIkrates, I lead production RAG and LLM-agent systems for pharma, banking, and legal, including document repositories of 20,000 to 125,000 documents. I built a document intelligence pipeline using hierarchical chunking (document → section → paragraph → sentence) and hybrid search to cut 15–20% failure rates versus pure semantic search.

I also drive retrieval quality end-to-end: metadata and table-first modeling with hybrid extraction, plus fine-tuning and serving (e.g., fine-tuned Qwen 32B with 20%+ accuracy gains). I’ve taken these systems to scale with quantization, vLLM serving, and multi-agent orchestration delivered as containerized ECS workloads behind API Gateway.