Looking for a job

Abdullah Kamal

@abdullahkamal

Message

I build production-grade LLM automations, RAG systems, and deployable AI agents for real workflows.

Pakistan

Message

What I'm looking for

I'm looking to build production AI automation, agentic systems, and RAG products where I can own work from discovery through deployment, collaborate with stakeholders, and keep improving systems based on real user feedback.

I've built and delivered production-style n8n automations for Fiverr and Upwork clients, including e-commerce tiering, lead scoring, support deflection, and community operations. I integrate OpenAI and Groq/LLaMA tool calling with reliable routing, retries, idempotency, and human-in-the-loop fallbacks.

My projects include a Dockerized Stripe documentation RAG API with hybrid FAISS and BM25 retrieval, an MCP server for Claude Desktop, and GitHub Actions deployment to AWS EC2. I've also built an autonomous code-fixing agent that verifies fixes with pytest and doctest, plus legal-contract and stock-market RAG applications using LangGraph, ChromaDB, and FastAPI.

I work end to end, from technical discovery and rapid prototypes through Docker, CI/CD, cloud deployment, monitoring, and iteration. Alongside AI automation, I've completed machine-learning work in risk modeling, customer segmentation, forecasting, and production-style data pipelines.

Experience

Work history, roles, and key accomplishments

Current

AI Automation Developer

Current

Fiverr / Upwork

Jan 2024 - Present (2 years 8 months)

Built and delivered production-style n8n workflow automations for clients, integrating LLM tool calling via Groq/LLaMA 3.3 and OpenAI for deterministic routing. Designed reliable integrations with retry/backoff, idempotency keys, and human-in-the-loop fallbacks.

N8n GROQ OpenAI Webhooks Gmail Slack Sheets Discord Retry Human In The Loop Make Zapier

Data Science Intern

DevelopersHub Corporation

Mar 2026 - Jun 2026 (3 months)

Completed end-to-end ML tasks including EDA, loan default risk with CatBoost, and customer segmentation with K-Means/PCA/t-SNE, packaging outputs into deployable artifacts. Built a Streamlit dashboard for non-technical stakeholders and ran small PEFT fine-tuning trials.

Python CatBoost K Means PCA T SNE Streamlit PEFT EDA