Auckland, New Zealand
Sarthak Rastogi
AI engineer building AI agents, RAG systems and ML platforms
Build Enterprise Agents with AgentCore (1)Agent Engineer - Senior AI Engineer Skills for Claude
AI Agent Engineering
My newsletter on building AI agents and RAG systems that hold up in production. Read by 16,500+ engineers and tech leaders.
About
I'm an AI Engineer with 3+ years of experience building AI agents and RAG systems.
I have two degrees in Data Science, and have done research work on ML in top labs worldwide.
Things I've built
- The central AI platform of an enterprise, where all complex agents in the company are built, bringing in $3M+ of cost saving.
- An AI agent system at a Microsoft-backed US startup, used by 200,000+ users a year.
Free AI training for non-technical leaders
80 short, plain-English lessons on why AI does what it does, written from what senior AI practitioners wish every leader understood. No technical background needed.
Experience
AI & Data Solution Engineer, One New Zealand
Feb 2025 – Present- Helped build the company's core agentic AI platform on AWS AgentCore. Built 12 MCP servers to integrate with external services. It underpins agents driving $2.6M+ in business value across 2 quarters.
- Built centralised gateways so other teams can use the platform's MCPs and agents in a governed, observable way.
- Built lineage-extraction algorithms to reverse-engineer logic across 30,000+ legacy SQL scripts, unlocking a $40M data modernisation program.
- Built a flagship customer-facing AI agent on Salesforce Agentforce.
Software Engineer, AI, Space and Time
Jun 2023 – Jan 2025- First AI hire at a $300M US startup. Built a natural-language-to-SQL agent processing 200,000+ queries a year.
- Co-architected and built DreamSpace, a multi-agent code-generation system.
- Delivered a text-to-SQL system on open-source LLMs for a multi-billion-dollar US legal firm, at 95% answer accuracy.
ML/AI Intern, Several AI labs and well-funded startups
2020 – 2023- Worked at Prodigal (YC Startup), AI Institute of South Carolina, TUM, IIT KGP, and more.
- Built a RAG-based natural-language-to-SQL system from scratch, reaching 93%+ accuracy within two weeks.
- Combined knowledge-graph retrieval with embedding models to build knowledge-infused embeddings.
- Integrated 200+ entity extractors into ProNotes for real-time phone-call processing.
- Trained transfer-learning models that calibrate low-cost air quality sensors, with an R² above 96%.
Open Source
Enterprise Agent System
An open-source platform for running AI agents for your company. Built on AWS Bedrock AgentCore. Managed entirely with Terraform.
Link to GitHub repoAgent Engineer
Turn your coding agent into a Senior AI Engineer - agent-engineer is a free and fully open - source plugin. It works on Claude Code, Codex, Cursor, etc. It contains skills with agent engineering knowledge (tools, prompting, context, RAG, evals, tracing, guardrails, production)
Link to GitHub repoRival AI
113Real-time prompt atack detection using custom lightweight models for production deployment
Link to GitHub repoSearch Expert
Search Expert uses a custom fine-tuned Small Language Model to understand natural language queries and extract only the fields explicitly mentioned — never hallucinating values that aren't there. It works across 10 search verticals out of the box. Built a year before Jev-like decision models.
Link to GitHub repoQualityPrompts AI
732Open-source Python library that applies 58 researched prompting techniques automatically. 700+ GitHub stars.
Link to GitHub repoKnowledge Graph RAG
272Automatically create knowledge graphs + document networks to boost performance on RAG
Link to GitHub repoEducation
IIT Madras
Programming, Data Science. Full ride for two years.
MRU
Computer Science, Data Science and ML. President's medal, full ride all four years.