Hi, I'm Pritam Mondal
Backend & Harness EngineerI build agentic AI systems with persistent memory, multi-agent pipelines, and production infrastructure. From memory layers backed by Redis + Neo4j to pitch deck generators with AI agents and background job queues.
- Agentic memory systems — Redis short-term rings, Neo4j knowledge graphs, and context window grounding for LLMs
- Multi-agent pipelines — BullMQ queues, NestJS orchestrators, and Mastra-powered AI agents with tool use
- Full-stack product engineering — Next.js frontends, Better Auth, Prisma databases, and background job processing
- Production infrastructure — Docker, CI/CD with CircleCI, Redis, RabbitMQ, and WebSockets
Tech Stack
The tools and platforms I use to build agentic systems, production backends, and full-stack products.
NestJS
Next.js
TypeScript
React
Mastra
Inngest
OpenAI SDK
Claude SDK
Redis
RabbitMQ
TanStack Query
WebSocket
Better Auth
Docker
AWS
CircleCI
Express.js
GitFeatured Projects
Real systems I've built — from memory-agentic layers to AI-powered product tools.
Address a memory. Get a grounded answer.
A query is a current through short-term cores, the knowledge graph, and the context window.
A memory agentic layer for LLMs — persistent memory across sessions with short-term Redis rings, long-term Neo4j knowledge graph, and a grounding context window.
- BFS algorithm — finds and merges duplicate memories in the knowledge graph
- Sliding window — keeps only the most relevant context for the AI
- Multi-agent pipeline — resolver, decision agent, then graph writer commits to Neo4j
Turn a startup idea into a pitch deck that raises.
Slides, visuals, and export-ready PDF or PPTX — in minutes.
Turn startup ideas into investor-ready pitch decks — slides, AI visuals, and export-ready PDF or PPTX in minutes.
- AI-generated pitch decks from a single prompt — no templates, no drag-and-drop
- YC-style 8-slide arc with AI-crafted visuals per slide
- Chat-based iteration and instant PDF/PPTX export
