Typebot integration
Brought the open-source Typebot chatbot builder into a monorepo, integrated it with the surrounding platform, and extended it with complex business logic.
Open to new roles · Remote / contract
Senior Software Engineer, AI · Zotok.ai
Explore ideas. Engineer solutions. Build systems.
I've spent 5+ years architecting full-stack applications and distributed systems. At Zotok.ai I lead a team building the LLM orchestration layer and scheduled agent workflows behind a multi-tenant B2B AI assistant.
alishahbaz617@gmail.com Hyderabad, India
01 / Experience
From payments and ERP integrations to browser automation, microservices, and LLM orchestration.
Senior Software Engineer — AI
Software Engineer
Software Engineer Intern
B.Tech, Computer Science
Brought the open-source Typebot chatbot builder into a monorepo, integrated it with the surrounding platform, and extended it with complex business logic.
Built backend services and APIs in NestJS while re-architecting a B2B commerce SaaS platform, alongside modular React/Next.js apps and reusable engineering foundations.
Integrated a desktop application with Tally ERP workflows, including the communication between the Electron client and backend services.
Architected a monorepo platform for independently developed apps: shared libraries, application boundaries, Module Federation, and build optimisation.
02 / Side projects
Open-source experiments in RAG, local-first sync, prompt optimisation, and backend infrastructure.
A deployed RAG chatbot that answers questions about my career in the first person. It grounds answers in my resume using hybrid BM25 + KNN search with cross-encoder reranking, and streams responses in real time.
An offline-first collaborative editor built on Yjs CRDTs and IndexedDB. Offline edits merge conflict-free on reconnect, and a Node.js WebSocket server with per-room state handles live cursors, presence, and typing indicators.
A GEPA-based prompt optimisation framework that tests LLM system prompts against real datasets, scores correctness, hallucination, format, and task completion, then iterates toward better prompts.
A background-job service. Jobs are persisted in Postgres with a status lifecycle, queued on Redis via Bull, processed concurrently per job type, retried with exponential backoff, and moved to a dead-letter queue when they keep failing.
03 / Toolkit
04 / Contact
Open to full-time roles, remote work, and contract engagements.