Building real-world AI systems, automating workflows, and shipping scalable products.
Powered by strong coffee and large language models βπ€
I'm KZ, a results-driven AI Tech Lead and Generative AI Developer focused on delivering intelligent, scalable, and business-aligned AI solutions. With 4.5+ years of deep tech experience, I build:
- πΉ End-to-end LLM applications with production-grade architecture
- πΉ AI scraping systems using GPT + Selenium + Scrapy
- πΉ High-performing PDF agents, chatbots, and RAG pipelines
- πΉ Custom SaaS platforms with smart automation baked in
I lead cross-functional teams, own delivery pipelines, and bring a laser focus on velocity + quality. Not here for theoryβhere to build.
- Python (FastAPI, LangChain, Streamlit, Gradio)
- JavaScript / TypeScript (React, Node.js)
- Shell Scripting, SQL
- Models: LLaMA 3, GPT-4, GPT-3.5, Mistral, Falcon, Claude
- Embeddings: OpenAI, HuggingFace, Groq
- RAG & Agents: LangChain, ChromaDB, Pinecone, Elasticsearch, FAISS
- Image & Video Gen: Diffusion Models, LoRA, Sora
- Tools: Selenium, Scrapy, BeautifulSoup, Playwright
- Pipelines: Auto-classification, data cleaning, structured output
- Databases: PostgreSQL, MongoDB, MySQL, Firebase
- Cloud: AWS S3, GCP, EC2
- DevOps: Docker, NGINX, IIS, GitHub Actions
- Streamlit, Plotly, Dash
- NLP-powered filters, dynamic graph generation
- Role-based dashboards & lead conversion panels
Here's what I'm cooking right now β from AI pipelines to production deployments:
- π§Ύ PDF Agent using LLaMA 3 + Groq with persistent vector indexing and contextual chat history
- π Multi-Document Query App with optimized RAG, Elasticsearch, and dynamic prompt engineering
- π A2A (Agent-to-Agent Communication): Coordinating autonomous agents for multi-step tasks
- π§ MCP (Multi-Component Pipeline): Architecting reusable modules for embeddings, retrieval, reasoning, and generation
- ποΈ Continuous session-based policy chatbots with session memory and source document tracking
- π¬ AI-powered customer support systems with feedback loops and relevance boosting
- π― Query understanding using GPT for smart routing (graph generation, lead filtering, etc.)
- β‘ Real-time AI scraping engines using Scrapy + GPT for classification & transformation
- π Auto-login scraping with Selenium/Playwright and storage to PostgreSQL + AWS S3
- π§Ή Smart document cleaning, tagging, and indexing pipeline
- π Custom lead analytics dashboards (Streamlit) with NLP-powered filters and graph insights
- π AI-powered search and trend analysis dashboards using vector databases
- π Role-based insights: Sales team breakdowns, stage-wise lead trends, conversion predictors
- π FastAPI + Streamlit apps with token-authenticated APIs, session handling, and user roles
- π‘ API development for internal tools and external client use (RAG, PDF QnA, chatbot endpoints)
- βοΈ Efficient background tasks & cron jobs for continuous processing and vector updates
- π§ Groq API integration with LLaMA3-70B for lightning-fast inference
- π¦ OpenAI, HuggingFace, Mistral support with LoRA fine-tuning for custom use-cases
- ποΈ ElasticSearch, FAISS, Pinecone pipelines with metadata-rich vector embedding storage
- π§ͺ Prompt engineering for better query disambiguation and agent accuracy
- π Memory systems for long-term context and multi-turn reasoning
- π₯ Generative media tools (text-to-video, image-gen pipelines using diffusion models)
πΌ LinkedIn: https://www.linkedin.com/in/kz2511/
π¬ Email: kunalzaveri11@gmail.com
π‘ Open to: GenAI consulting, contract roles, startup collabs, and tech leadership gigs.
βI donβt just prompt LLMsβI prompt outcomes.β
β Explore my repositories, drop a star, and letβs build the next-gen AI stack together.