class DevDoshi:
role = "AI Engineer & LLM Systems Builder"
degree = "B.Tech — Artificial Intelligence & Data Science"
cgpa = 9.48
institute = "ADIT, CVM University"
focus = ["Generative AI", "Agentic Workflows", "MCP Servers"]
stack = ["LangChain", "LangGraph", "LangSmith", "FastMCP", "FastAPI"]
learning = ["Deep Learning", "DSA", "RAG Optimization"]
motto = "Build real things. Learn in public. Ship often."AI & Data Science undergraduate with a 9.48 CGPA, specializing in LLM applications, multi-agent systems, and automation pipelines. I believe in learning by building every repository here is a production-grade system, not a tutorial project.
I specialize in LLM orchestration, MCP server development, and agentic pipeline engineering — designing intelligent, autonomous workflows with LangGraph, LangSmith, and FastMCP. Currently building at the intersection of real-world AI deployment and observable, maintainable system design.
Open To
Languages
Backend & APIs
ML / DL Frameworks
LLM / GenAI Stack
Vector Stores & Automation
Tools
| Domain | Proficiency | Details |
|---|---|---|
| LLM Orchestration | ████████░░ Advanced |
LangChain, LCEL, multi-step agentic pipelines |
| Agentic AI Systems | ████████░░ Advanced |
LangGraph state machines, n8n multi-agent routing |
| MCP Server Development | ████████░░ Advanced |
FastMCP typed tools, async SQLite, cloud deployment |
| LLM Observability | ███████░░░ Intermediate |
LangSmith tracing, evaluation, prompt versioning |
| Generative AI (RAG) | ████████░░ Advanced |
RAG pipelines, vector stores, retrieval tuning |
| Multi-LLM Integration | ███████░░░ Intermediate |
Gemini, OpenAI, Groq, Ollama, Claude — unified APIs |
| Deep Learning | ██████░░░░ Intermediate |
TensorFlow, ANN, CNN — supervised & unsupervised |
| Automation Pipelines | ███████░░░ Intermediate |
n8n workflows, Google Sheets API, scheduled agents |
AI Engineer · AIGyde
May 2026 — Present
Designed and deployed a production-grade AI News Creation Pipeline that autonomously monitors, summarizes, and delivers the latest news to mail fully automated, zero manual intervention.
- Built an end-to-end news aggregation and summarization system using LangChain and LLM APIs, covering events from the past 24 hours
- Architected a scheduled email delivery engine that triggers every 3 hours, sourcing fresh news via NewsAPI and formatting with Jinja2 templates
- Implemented smart caching and rate-limiting layers to minimize API overhead across high-frequency scheduled runs
- Integrated LangSmith observability for monitoring summarization quality and iterating on prompt performance in production
| Recognition | Details |
|---|---|
| 🏆 Smart India Hackathon | Participated · 2024 & 2025 |
| 🎯 College Ideathon | Participated · 2024 & 2025 |
| 📈 Academic Excellence | CGPA 9.48 · B.Tech AI & Data Science · ADIT, CVM University |
| 🚀 Production AI Builder | Multiple end-to-end LLM and MCP systems deployed and live |
SAP · Edunet Foundation
Python Programming · Data Analysis · Artificial Intelligence · SAP Conversational AI Chatbot — ADIT, CVM University · 2024–2025 · Certificate ID: CU25_18349
HackerRank
"The best way to understand intelligent systems is to build one — imperfectly, iteratively, in public."