Mohammad Ayyub
Freelance Agentic AI Development | Custom AI Agents for Scaling Businesses | AI/ML Developer
Raipur
3+ years of work experience
Jun 20251 yr 5 mosPresent
United States
Designing and delivering production-ready AI solutions across Agentic AI, LLMs, RAG, Voice AI, workflow automation, and Generative AI for startups and businesses.
Key Contributions
• Built AI agents, multi-agent systems, and business automation workflows using LangGraph, CrewAI, n8n, MCP, and Python.
• Developed enterprise RAG document intelligence platforms capable of retrieving information from 5M+ enterprise records using Qdrant, semantic search, hybrid retrieval, reranking, and citation-backed responses.
• Engineered a YouTube Transcript RAG Chatbot enabling contextual conversations over video transcripts with hybrid retrieval, query expansion, reranking, and grounded answers.
• Built AI-powered Voice Agents using VAPI, LiveKit, and n8n for lead qualification, appointment booking, and automated customer interactions.
• Developed an AI-powered Photo Memory Platform featuring contextual retrieval, face recognition, vector embeddings, semantic search, intelligent memory organization, and LLM-powered search.
• Designed REST APIs, AI orchestration pipelines, and production-ready backend services using Python, FastAPI, Docker, PostgreSQL, Redis, and cloud infrastructure.
• Built LLM applications using OpenAI, Claude, Gemini, and open-source models, integrating external APIs, business tools, and knowledge bases into scalable AI workflows.
• Optimized Generative AI inference pipelines, reducing GPU memory requirements and improving deployment efficiency for production environments.
Core Technologies
Python • FastAPI • LangGraph • CrewAI • LangChain • LlamaIndex • RAG • Qdrant • ChromaDB • OpenAI • Claude • Gemini • VAPI • LiveKit • n8n • Docker • PostgreSQL • Redis • PyTorch • TensorFlow
Jan 20233 yrs 10 mosPresent
United States
I architect Generative AI/ML/DL development with a focus on resource efficiency and seamless integration.
Key Achievements & Deliverables:
Infrastructure Optimization: Successfully reduced VRAM usage for high-demand diffusion models from 50GB+ to 16GB, achieving a 60%+ increase in hardware efficiency for production scaling.
Model Development: Fine-tuned custom LLMs and image pipelines tailored for specific hardware constraints (VRAM, CPU, Load).
Advanced Prompt Engineering: Solved 77-token limitations and enabled scheduled prompt parsing to improve model output quality and reliability.
AI Automation: Deployed production-grade autonomous voice agents using n8n,VAPI and GHL to streamline business workflows.
Tech Stack: Stable Diffusion, LoRA, LLMs, n8n, Vapi, Python, PyTorch.
Jan 20256 mosJun 2025
Seoul, South Korea
1. AI Model Development and Research: Developed and optimized AI models including Stable Diffusion, Img2Img, Img2Vid, Txt2Vid, and Neural Style Transfer (NST) for artistic content generation using IP-Adapter, ControlNet, and Textual Inversion.
2. Data Collection and Preprocessing: Curated and preprocessed high-quality datasets using Selenium-based web scraping and open-source repositories for AI model training.
3. Fine-Tuning AI Models: Fine-tuned and optimized LLM, Diffusion, Image, and Video models using LoRA, QLoRA, and PEFT to enhance output quality and meet project specifications.
4. Documentation and Reporting: Created comprehensive technical documentation and delivered weekly progress reports on development activities.
5. Service Integration: Integrated AI models into web platforms via API development and collaborated with UI/UX teams for seamless deployment.
6. Market Research and Trend Analysis: Conducted competitive analysis and trend monitoring to identify strategic opportunities in AI technologies.
7. Technical Support for Projects: Provided remote technical support for AI model deployment during exhibitions and client projects.
8. Model Optimization & Prompt Engineering: Optimized LLM and Diffusion models for maximum VRAM, CPU, and inference efficiency; engineered prompts to resolve 77-token limitations, enable prompt weighting, and implement scheduled prompt parsing.
9. Innovation and Idea Contribution: Drove innovation through strategic ideation and participation in brainstorming sessions to enhance company offerings.
Feb 20244 mosMay 2024
New raipur
Responsible for designing, building, and maintaining full-stack web applications using Python and the Django framework. Worked across the entire development lifecycle — from database architecture and server-side logic to frontend integration and production deployment.
Developed and deployed multi-module web applications using Django (MTV architecture), handling views, URL routing, templates, and ORM queries.
Designed and managed relational database schemas in MySQL; wrote optimised queries and handled migrations using Django migrations framework.
Implemented secure Google OAuth 2.0 authentication via Google Cloud Console, including OAuth consent screen setup and redirect URI configuration.
Built and consumed RESTful APIs using Django REST Framework for frontend-backend communication and third-party integrations.
Configured environment variable management across Windows and macOS/Linux for secure handling of credentials (DB, OAuth keys).
Created structured technical documentation for local deployment covering virtualenv setup, DB import, and Google Cloud configuration for teams and beginners.
Applied Python virtual environments and requirements.txt for reproducible dependency management across development environments.
Participated in full-lifecycle delivery: requirements gathering, system design, development, testing, and deployment handoff.
Mohammad’s AI Stack
Education
Master's Degree, Computer Programming, Specific Applications
Mats university
20222 yrs2024
Master of Computer Applications - MCA, Computer Science
Mats university
20222 yrs2024
Join Mohammad and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal Boss