Machine Learning Engineer & Agentic AI Specialist
I build intelligent agents that plan, reason, and act — from multi-agent research systems and RAG pipelines to agentic coding workflows with Claude Code and Kimi Code.
I'm a Machine Learning Engineer from Pakpattan, Pakistan, focused on the frontier of agentic AI — systems where LLMs plan, use tools, and collaborate to complete real work end to end.
My day-to-day toolkit includes Claude Code and Kimi Code for agentic software development, LangGraph for multi-agent orchestration, and deep-agent harness patterns (planning, subagents, filesystem memory) for long-horizon tasks.
I hold a BS in Computer Science from The Islamia University of Bahawalpur, specializing in Machine Learning and Artificial Intelligence.
Hands-on with the current wave of agentic AI tools, models, and protocols.
CLI coding agents that read codebases, edit files, run tests, and manage git autonomously. I use both daily to ship software faster and teach teams how to harness them.
Harness engineering for long-horizon agents: a prime agent that plans, writes todos, delegates to subagents, and persists state to a filesystem backend — the pattern behind modern deep-agent frameworks.
Orchestrating planner, researcher, and writer agents with shared state, tool use, and human-in-the-loop checkpoints for research and automation workflows.
Working with Kimi K2, DeepSeek, and Qwen — open models with strong reasoning, tool-calling, and agentic capabilities, deployable without frontier-API lock-in.
Connecting agents to tools and data via the Model Context Protocol, and agent-to-agent communication for interoperable, composable AI systems.
Retrieval-augmented generation with vector and graph stores (Pinecone, Neo4j), plus context engineering — memory, compaction, and prompt design that keep agents reliable.
Voice platform using Twilio and custom LLM agents with real-time speech recognition, appointment scheduling, and automated follow-ups — 500+ calls daily, 95% customer satisfaction.
No-code platform for building AI-powered business automations with drag-and-drop agent blocks and triggers — integrates with 50+ apps and removes hours of repetitive work per team weekly.
LangGraph-based system where planner, researcher, analyzer, and writer agents collaborate with shared memory and tool use — compresses hours of desk research into minutes.
Swarm of specialized review agents — security analyzer, performance optimizer, documentation checker — that collaborate on PR reviews and catch issues before merge.
Self-healing ETL pipelines whose agents detect errors, plan fixes, and adapt to schema changes on their own — reduced data processing time by 70%.
Deep learning system for early diabetic retinopathy screening from retinal images — clinical decision support that helps catch vision-threatening disease sooner.
Chat-with-your-documents app built on retrieval-augmented generation — upload a PDF and ask questions with cited, grounded answers.
Reinforcement learning trading agent using PPO with custom reward shaping, market state encoding, and risk constraints — 23% annual returns in backtesting.
Helping Hand
Advising on AI strategy and deploying LLM-powered solutions — chatbots, RAG systems, and workflow automation — that cut manual workload and improve response quality.
NAVTTC (National Vocational & Technical Training Commission)
Coordinating government-funded AI/ML training programs — curriculum planning, instructor management, and hands-on labs covering Python, ML, and modern agentic AI tools for hundreds of trainees.
Aridian Technologies · Rawalpindi
Building production ML and agentic AI systems for clients.
Essenceware Technologies · Rawalpindi
Teaching ML, deep learning, and agentic AI to engineering teams.
Neuroimaging Research Group, Air University · Islamabad
Applied deep learning to neuroimaging analysis and classification.
Sparking Asia · Rawalpindi
Data pipelines, dashboards, and analytics for business decision-making.
Open to AI projects, collaborations, and consulting on agentic systems.