Skip to content
View AhmedIrfan7's full-sized avatar
10 followers · 21 following

Achievements

Achievement: Pull SharkAchievement: QuickdrawAchievement: YOLO

Achievements

Achievement: Pull SharkAchievement: QuickdrawAchievement: YOLO

Highlights

  • Pro

Block or report AhmedIrfan7

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
AhmedIrfan7/README.md
Typing SVG


BSCS @ FAST NUCES, Islamabad · AI & Automation intern, Jul–Sep 2026 · open to freelance & full-time work

━━━━━━━━━   if it doesn't deploy, it doesn't count   ━━━━━━━━━


01   the pipeline

Everything below is one shape, repeated: an idea goes in, agents reason over it, and something a stranger can click comes out the other end.

REASON
design the graph




GROUND
keep it honest




ORCHESTRATE
wire it together




SHIP
make it real






02   systems, running

Every metric here is real, pulled from the repo's own README and release history, not decoration.


AgentForge  ·  v1.0.0 tagged
An open-source, multi-tenant AI platform where an organization uploads its knowledge, configures specialized agents, and deploys them as chat and voice assistants behind one embeddable widget. Real tenant isolation, layered memory, explainable retrieval, and a multi-agent orchestrator, not a wrapper around one LLM call.

public roadmap 300/300 steps
CI 7 workflows, CodeQL + audits
FastAPI RAG Postgres Docker
 

File Atlas  ·  v1.0.0 released
A local-first desktop app in Rust that indexes every file you own and helps you reshape it safely: sub-100ms search, BLAKE3 duplicate detection, a storage treemap, and natural-language search over your own machine with a local model. Cloud is opt-in, per request, never silent.

platforms Windows · macOS · Linux
data on-device by default
Rust Tauri TypeScript Ollama
 

Dental Clinic RAG Voice Assistant
A voice-and-text AI receptionist as a 10-workflow multi-agent system in n8n. Give it a clinic URL, it scrapes, extracts facts, then fans a query out to pricing / booking / FAQ agents in parallel and merges the replies.

hallucination evals 14/14 pass
workflows 10, authenticated
n8n Qdrant Postgres Voice
 

Multi-Agent Research Assistant
Four LangGraph agents, Clarity, Research, Validator, Synthesis, that route between each other on confidence scores and pause to ask when a question is ambiguous, before streaming the answer word by word.

coordinated agents 4
response delivery SSE, live
LangGraph FastAPI React 19 Anthropic
 

PEERLESS.AI  ·  National AI Hackathon '26
Seven agents that surface research-integrity concerns in scientific papers: GRIM checks, statcheck p-value recomputation, DOI verification against Crossref. Findings only, never verdicts.

specialized agents 7
review model human-in-the-loop
Streamlit Groq LLaMA-3.3 Crossref
 

ResumeIQ  ·  live demo ↗
A two-stage pipeline extracts a structured candidate profile, scores it against the role, and hands HR an AI-drafted reply that never sends without a human click.

pipeline extract → score
deployment live on Vercel
n8n Next.js OpenAI Gmail API
 

German Driving-School Chatbot  ·  fine-tune ↗
A bilingual pricing bot built on one rule: it must never guess a price. Every number is read verbatim from a verified store and cross-checked against the source PDF. Paired with a QLoRA fine-tune of Mistral-7B on the same domain, exported to GGUF to run locally.

price accuracy 100%
test suite 80, all passing
RAG QLoRA Docker DE / EN


03   upstream

Work that landed in other people's codebases, reviewed by their maintainers.

Project Contribution Status
jamiepine/voicebox
open-source AI voice studio		 #895  fix(mcp)   added the missing model_size parameter to voicebox.speak, so agents can pick a model instead of silently getting the default merged
grayhatdevelopers/vidxp
natural-language video indexing engine		 #144  feat(search)   made retrieval and fusion scores self-describing, so a ranked result explains why it ranked open


04   the automation line
Lead Generation System
live dashboard ↗		 AI Support Router RAG Pipeline Chatbot
A sheet row becomes a scraped, redesigned, deployed preview site. Zero manual steps, idempotent by construction. A local model classifies incoming Gmail and routes it to the right team. Watches a Drive folder, indexes new docs into Pinecone, answers questions over them.
n8n GitHub API n8n Local LLM n8n Pinecone

 :file_folder: everything else in the repo
Project What it is
UrbanFix · live ↗ Civic issue tracker. Citizens pin problems on a live map, admins manage them from a dashboard
RACE QuizEngine Reading-comprehension quiz generator on the RACE dataset. No LLM, just an LR/SVM/NB ensemble
LinkedIn Content Creator Sheets topics → Tavily research → AI-drafted posts, queued automatically
Customer Support Workflow Classifies and routes support email end to end
Super Mario Bros. in x86 Assembly Mario rebuilt at the register level: physics, collisions, day/night cycle
SmartCity Simulation Custom graphs, heaps, hash tables, Dijkstra routing over a city map
Sonic Classic Heroes Momentum-based Sonic physics engine in C++ and SFML


05   the range

most people who build agents have never written a line of assembly, and most people who write Rust have never shipped an agent graph. I move the whole stack.





06   open a channel

building with agents, RAG, or automation, or want the whole pipeline handled? I'd like to hear about it.


Pinned Loading

  1. multi-agent-research-assistant multi-agent-research-assistant Public

    Full-stack AI research assistant with 4 specialized LangGraph agents, live streaming responses, human-in-the-loop clarification, and multi-turn conversation. Built with FastAPI and React.

    Python

  2. n8n_dental_rag_orchestration n8n_dental_rag_orchestration Public

    n8n multi-agent RAG voice + chat assistant for a dental clinic — parallel pricing/booking/FAQ agents merged into one answer, Qdrant + Postgres, 14/14 hallucination-probe evals passed

    JavaScript 1

  3. hr_resume_screening hr_resume_screening Public

    ResumeIQ — two-stage AI resume screener (extraction + rubric scoring) with an AI email assistant for HR outreach. n8n + Next.js + OpenAI + Gmail API, deployed live on Vercel

    TypeScript 1

  4. LeadGenerationSystem LeadGenerationSystem Public

    Fully automated n8n pipeline: a prospect row becomes a redesigned, deployed preview site — scrape to OpenAI spec to built files to GitHub Pages to an emailed live URL, tracked on a dashboard

    Python

  5. RagPipelineChatBot-GermanDataset RagPipelineChatBot-GermanDataset Public

    Bilingual DE/EN driving-school chatbot: deterministic price engine (prices never LLM-generated) + hybrid BM25/embeddings RAG in Python/FastAPI, with a native n8n multi-agent orchestration layer

    Python 1