I design and develop intelligent systems, scalable backend architectures, and high-performance cross-platform applications. My work integrates advanced AI capabilities with clean, reliable engineering to build practical, production-grade solutions.
- Building production-grade agentic AI systems and Odoo / ERP integrations
- Publishing open-source agent skills for coding agents like Claude Code, Codex and Antigravity
- Portfolio: portfolio-anas-lari.vercel.app
- LLM Systems: Development, fine-tuning, and prompt optimization.
- Computer Vision: Face recognition, document analysis, forgery detection, and OCR pipelines.
- Reinforcement Learning: Custom training environments and intelligent agent design.
- Performance: Multimodal reasoning engines, dataset creation, and model quantization.
- Agentic Systems: Autonomous, tool-using agents built for real business workflows.
- Odoo / ERP: Module development and AI assistants embedded in ERP systems.
- Agent Skills: Reusable skill packages that extend coding agents.
- Architecture: Microservices, distributed systems, and real-time WebSocket communication.
- Backend APIs: Robust service layers with FastAPI, Node.js/Express, and Laravel.
- Frontend UI/UX: Component-driven, responsive applications with React and Next.js.
- Security & Data: Secure JWT/OAuth2 authentication, task queues, and data pipeline management.
- Frameworks: Production-grade Flutter apps using BLoC, Provider, and Riverpod.
- Cloud Integration: Firebase Authentication, Firestore, and scalable Cloud Functions.
- Experience: Custom animations, responsive layouts, and native platform integrations.
- Data Stores: PostgreSQL, MySQL, Redis, and Firestore.
- Containerization: Dockerized backend and AI workflows.
- CI/CD: Automated build, test, and deploy pipelines using GitHub Actions.
- Infrastructure: Linux server configuration, deployment, and systems administration.
| Project | What it is | Stack |
|---|---|---|
| max-ai | Full-stack multimodal personal assistant with a voice-first UX, real-time wake-word activation, face authentication, retrieval-augmented intelligence and a 3D hologram interface. | Python |
| SHIAN-OS | A personal AI operating system that unifies voice intelligence, real-time perception and automation. | C |
| Multi-Modal-Health-AI-Agent-Guide | MM-HIE Agent: fuses symptoms (NLP), labs (OCR), medical imaging and vitals for explainable differential diagnosis, risk scoring and clinical reports. | Python |
| kolkata-tour-ai | Offline-capable, location-aware AI travel assistant for exploring Kolkata. | Dart, Flutter |
| air-keyboard | Experimental virtual keyboard: type in the air using only your hands and a standard webcam. | Python |
Open-source skills for coding agents (Claude Code, Codex, Antigravity and others).
| Skill | What it does |
|---|---|
| job-hunt-outreach | Runs a full job hunt from your CV: finds openings, verifies them, builds a spreadsheet, applies and drafts cold emails with your approval. |
| verify-before-done | Makes the agent run and verify any code or config change before it reports the task as done. |
| inventor-mindset | Pushes the agent past the obvious solution when building something new, with first-principles thinking. |


