I build practical AI-powered software, automation systems, and developer tools using Python, n8n, LLM APIs, webhooks, and backend technologies.
I learn by building real systems — designing workflows, connecting APIs, testing real use cases, debugging failures, and shipping working software.
Currently building Tukdify Labs, an independent software studio focused on local-first software, open-source AI tools, and self-hosted workflows.
Build → Test → Break → Debug → Ship
Tukdify Labs is my independent software studio focused on building practical software that gives users more control.
The current focus includes:
- Local-first software
- Open-source AI tools
- Self-hosted automation workflows
- Practical developer utilities
- Software designed to work without unnecessary cloud lock-in
| Project | What it does | Technologies |
|---|---|---|
| LeadIQ-AI | AI-powered lead qualification, deduplication, scoring, and multi-channel routing workflow. | Python n8n Gemini PostgreSQL |
| Tukdify Clips | Local-first AI tool for turning long-form videos into short vertical clips with transcription, segment selection, reframing, and captions. | Python Whisper FFmpeg |
| Tukdify MultiCompressor | Offline batch compression for video, image, and audio files with configurable quality presets. | Python FFmpeg Pillow |
| Tukdify Video Downloader | Desktop media downloader for high-resolution video and audio extraction with resumable downloads. | Python yt-dlp FFmpeg |
| AI-Tool-Research-Lab | Structured research and evaluation of AI tools, workflows, experiments, and practical use cases. | Python LLMs Automation |
My main engineering work is around turning AI capabilities into useful, reliable workflows.
- LLM API integration
- Structured outputs
- Prompt and system architecture
- Classification and intent detection
- AI-assisted decision workflows
- AI agent experiments
- n8n
- Webhook-driven workflows
- API orchestration
- Email automation
- Lead qualification and routing
- Multi-channel notifications
- Database and service integrations
- Python development
- REST API integration
- Backend scripting
- Data processing
- Local file processing
- Desktop software
- Automation tooling
Python · JavaScript · SQL
n8n · LLM APIs · Google Gemini API · AI Agents · Prompt Engineering
REST APIs · PostgreSQL · Supabase
FFmpeg · Pillow · Whisper · Desktop GUI Development
Git · GitHub · Docker · Linux · Bash · VS Code
I prefer solving a real problem with a working system over building another AI demo.
LLMs are useful for reasoning, classification, and ambiguity.
Critical execution should still be handled with code, validation, explicit logic, and predictable failure handling.
Software should give users meaningful control over their own data, files, and execution environment whenever the problem allows it.
APIs fail. Inputs are incomplete. Networks disconnect. Models behave unexpectedly.
Good systems should make those failure modes visible, recoverable, and understandable.
- Building AI-powered automation systems
- Developing local-first software
- AI agents and practical LLM workflows
- API and backend engineering
- Open-source software
- Shipping and improving Tukdify products
- Turning real problems into reusable systems
- AI agent architectures
- Production LLM systems
- MCP
- Distributed systems
- Advanced backend engineering
- Reliable automation infrastructure
🌐 Tukdify Labs: tukdify.com
💼 LinkedIn: Sourabh Jangid
📧 Email: sourabhjangid.dev@gmail.com