Maharshi Shastri

I’m a Full Stack AI Engineer focused on building autonomous AI systems with task planning, RAG pipelines, streaming architectures, and human-in-the-loop execution workflows. I enjoy designing end-to-end AI products that turn model capabilities into real, usable systems.

In my recent Chronos-AI work, I built a full-stack autonomous AI platform combining conversational AI, task planning, RAG memory systems, and human-in-the-loop execution workflows. I designed a multi-queue AI orchestration architecture using Redis + Celery, and implemented semantic RAG retrieval with grouped chunking, metadata-aware indexing, cross-encoder reranking, and token-budgeted context summarization.

I also developed streaming telemetry infrastructure using SSE for real-time mission execution, live KPI monitoring, and checkpoint-based mission continuation. I engineered tool-routing agents with browser automation and sandboxed Python execution integrated into planner-executor workflows.

Before that, I built Python-based workflow automation for PDF financial statements using MongoDB and document parsing pipelines, including rule-based highlighting for analytical patterns and KPI-oriented analysis pipelines under NDA constraints. I’ve also developed ML pipelines for structured-data prediction tasks (diabetes risk analysis, fraud detection, cancer diagnosis, and clustering-based segmentation) from preprocessing and feature engineering through model training, evaluation, and experimentation with Python-based tooling.