AutoDev Agent Team
A multi-agent system using LangGraph to autonomously generate, test, and document full-stack apps. Features a Supervisor-Worker architecture and an E2B sandbox based self-debug loop.
> INITIALIZING AGENTIC CORE... _
SYSTEM OFFLINE
Architecting Autonomous Minds & Future Logic.
I build Multi-Agent Systems where LLMs don't just chat—they execute, reason, and code reliably.
The Engineering Philosophy
From Vectors to Systems
My background in Mathematics (B.Sc) gave me a strong grasp of Linear Algebra. While others see AI output as magic, I understand the underlying Vector Embeddings and Matrix Operations that power semantic search and RAG systems.
I transitioned into MCA to translate this theoretical intuition into production-grade software. Today, I don't just use Vector Databases; I understand the math that makes them work, allowing me to build more accurate Agentic Systems.
Core Engineering Tenets
🌊
Flow Engineering
Moving beyond simple "Prompt Engineering". I design iterative loops where agents plan, execute, critique, and refine their work.
🧠
System 2 Reasoning
Forcing models to "think before they speak." Using Chain-of-Thought (CoT) and reflection steps to solve complex logic puzzles.
📉
Eval-Driven Dev
You can't improve what you don't measure. I build custom evaluation pipelines to benchmark agent performance.
How I Build
01
Deconstruct
I treat complex AI problems like mathematical proofs. I break vague requirements down into atomic components.
02
Orchestrate
Design Multi-Agent Systems with clear roles (Supervisor, Coder, Reviewer) and strict Graph flows.
03
Optimize
Leveraging tools like Groq LPU for speed and Docker for consistent environments.
The Journey
2024 - Present | Master of Computer Applications (MCA)
Specializing in Agentic AI
Started building autonomous software engineers using LangGraph. Deep diving into GraphRAG and LLM Orchestration. Building "AutoDev" to automate coding workflows.
2021 - 2024 | B.Sc in Mathematics
The Foundation
Graduated with a focus on abstract logic and rigorous problem-solving. Developed the analytical mindset required for understanding Deep Learning algorithms.
Goal
AI Systems Engineer
Seeking a role to push the boundaries of what Autonomous Agents can do in production environments. Ready to debug, deploy, and scale.
Technical Arsenal
LangGraph
Supervisor-Worker Patterns & Cyclic Graphs.
Neo4j & Chroma
Hybrid Memory Systems (Graph + Vector).
Groq & LLM Ops
Sub-second inference & Fine-tuning (Unsloth).
FastAPI & E2B
Async Backends & Code Execution Sandboxes.
Deployed Systems
Unified Medical RAG Agent
A unified agent pipeline on Hugging Face Spaces combining ChromaDB and Tavily Search. Solved read-only filesystem issues using UUID-based storage.
Air Quality Index System
A low-level system developed in C to monitor real-time AQI. Implemented HTTP data retrieval using cURL commands to fetch environmental data efficiently.
Impact & Learning
🎛️
LLM Fine-Tuning
Successfully fine-tuned Llama-3 & Gemma models using Unsloth. Optimized models for specific "Text-to-JSON" tasks to enhance agentic reliability.
🛠️
Open Source Builder
Building public-in-public. My AutoDev Agent repository demonstrates the capability to architect complex, multi-file systems from scratch.
🧠
Deep Learning Path
Solid conceptual understanding of Deep Learning (CNNs, RNNs). Currently actively upskilling in PyTorch to translate theory into code.