Abdul Sammad
Artificial Intelligence Engineer
Profile in progress
Abdul is building their profile!
Just built my automation on n8n.io (http://n8n.io)
for the first time after working on make.com (http://make.com)
, and I'm not going back to manual work ever again.
Used to update cricket stats by hand after every match. Runs, balls, wickets, sixes — copy-pasting from a PDF into Google Sheets like it's 2005.
Now I just drop the scorecard PDF in a folder and walk away.
n8n detects it → Claude AI reads it → Google Sheets updates itself.
Batting stats. Bowling stats. All players. Zero clicks.
Let's go.
hashtag#n8n (https://www.linkedin.com/search/results/all/?keywords=%23n8n&origin=HASH_TAG_FROM_FEED) hashtag#AIAutomation (https://www.linkedin.com/search/results/all/?keywords=%23aiautomation&origin=HASH_TAG_FROM_FEED) hashtag#LowCode (https://www.linkedin.com/search/results/all/?keywords=%23lowcode&origin=HASH_TAG_FROM_FEED) hashtag#ClaudeAI (https://www.linkedin.com/search/results/all/?keywords=%23claudeai&origin=HASH_TAG_FROM_FEED)
0
3
Certified in Clude with Anthropic API course with hands-on experience in deploying multiple Ai Automations and Projects using latest cutting-edge technolgies and tools like LangGraph, RAG, MCP, n8n, Make
0
9
Just built a self-correcting research agent that actually knows when it doesn't have enough information and researches until it does.
DeepSearch Agent is a LangGraph-powered research tool that breaks down your question, searches 3 sources in parallel (web, Wikipedia, arXiv), grades every result for relevance, and synthesizes a cited markdown answer.
The key insight: a self-correction loop. After the first search pass, a grader LLM scores every result 0.0-1.0. If fewer than 2 high-quality results are found, the agent rewrites the query and re-searches up to 2 retries. No hallucinated confidence. Real quality gates.
Tech stack:
LangGraph for the graph orchestration (parallel Send, conditional edges, typed state)
Groq (llama-3.3-70b-versatile) for free, fast LLM inference
Tavily for real-time web search
ArXiv for technical papers
Wikipedia for background context
LangSmith for live tracing of every step
LangGraph Studio for the visual graph runner
Architecture: query_analyzer (decomposes question) -> parallel search (web + wiki + arXiv) -> relevance_grader (scores + decides retry) -> synthesizer (cited markdown answer) -> END
The agent runs in interactive mode so you can ask follow-up questions in the same session. Every run is traced in LangSmith so you can see exactly which node triggered the retry loop, how many tokens were used, and where time was spent.
Built with clean, interview-ready code. The kind of project you can walk a recruiter through node by node.
hashtag#LangGraph (https://www.linkedin.com/search/results/all/?keywords=%23langgraph&origin=HASH_TAG_FROM_FEED) hashtag#LLM (https://www.linkedin.com/search/results/all/?keywords=%23llm&origin=HASH_TAG_FROM_FEED) hashtag#Agent (https://www.linkedin.com/search/results/all/?keywords=%23agent&origin=HASH_TAG_FROM_FEED) hashtag#Python (https://www.linkedin.com/search/results/all/?keywords=%23python&origin=HASH_TAG_FROM_FEED) hashtag#OpenSource (https://www.linkedin.com/search/results/all/?keywords=%23opensource&origin=HASH_TAG_FROM_FEED) hashtag#Groq (https://www.linkedin.com/search/results/all/?keywords=%23groq&origin=HASH_TAG_FROM_FEED) hashtag#AIResearch (https://www.linkedin.com/search/results/all/?keywords=%23airesearch&origin=HASH_TAG_FROM_FEED)
0
11
🚀 Just built something I'm really proud of: SupportMind, a multi-agent AI customer support system using LangGraph + Groq.
The idea was simple: instead of one AI trying to do everything, why not build a team of agents? A supervisor reads the ticket, figures out what the customer needs, and sends it to the right specialist: FAQ, billing, or technical.
The coolest part? Human-in-the-Loop. When a refund request comes in over $100, the entire graph just stops and waits for a manager to approve it. No hacks, no workarounds. LangGraph's interrupt() handles it natively. The workflow literally pauses until a human says yes or no.
A few other things I built into it: 🔀 Smart routing with escalation (agents can hand off to each other) 🧠 Auto memory compression so token costs stay flat no matter how long the chat gets 💾 Sessions that survive restarts using SQLite 📊 Full graph tracing in LangGraph Studio
Powered by Groq's Llama 3 70B. The speed is insane, responses feel real-time.
Drop a comment or DM me if you want the code! 👇
0
17