HEY 👋 , I'M
Asgar Ali Sha D
GenAI Developer | LLM & RAG Engineer | Agentic AI
GenAI Developer with 5+ years of software engineering experience, the last two focused on shipping LLM-powered systems to production. I have built RAG pipelines over vector stores, an agentic tool layer of 3 MCP servers giving non-technical users natural-language access to SQL, documents, and repositories, and prompt chains extracting structured data from 100+ document formats at 95%+ accuracy — automating 80% of a manual data entry workflow. I work across the full delivery path: prompt and retrieval design, Python service implementation with FastAPI and LangChain, and containerised deployment through CI/CD.
About Me
I design and ship AI systems that reach real users: retrieval pipelines, LLM extraction services, Model Context Protocol servers, and the FastAPI, Django, and React applications that surround them. Five years of full-stack and enterprise integration work sit underneath the AI — so the models land in software that holds up in production.
Currently working as Specialist – Software Engineering at
InteleANTS Virtual Workforce Pvt Ltd
Aug 2024 - PresentCurrent
Chennai, Tamil Nadu GenAI & Full Stack Engineering
- Built an AI customer support chatbot with a RAG knowledge-retrieval pipeline over a vector store, serving multi-channel support through Chatwoot webhook integration, with automated escalation to human agents when retrieval confidence or intent fell outside supported scope.
- Developed 3 MCP servers with FastMCP exposing tool-use interfaces for read-only SQL exploration and dependency tracing, hybrid semantic search over document collections with relevance thresholding and pagination, and Azure Git and work-item analysis — letting non-technical users query enterprise data in natural language.
- Designed prompt chains for structured data extraction reaching 95%+ accuracy through few-shot prompting, chain-of-thought reasoning, and Pydantic schema validation with automatic retry on malformed output, across 100+ heterogeneous document formats.
- Integrated LLMs (OpenAI, Ollama) via LangChain into production ETL pipelines for factory PO processing, firewall data extraction, and vendor data synchronisation, automating 80% of manual data entry and routing sensitive workloads to self-hosted models to control cost and data exposure.
- Shipped these systems as production Python services: a full-stack Call Audit QA platform (FastAPI, SQL Server, React) with 25+ features adopted by 30+ auditors, TensorFlow transformer-embedding models for call sentiment classification, and PySpark feature pipelines over millions of call records — containerised with Docker and released through Azure DevOps CI/CD.
Codentrix Technologies Pvt Ltd
Sep 2021 - Jul 2024
Chennai, Tamil Nadu codentrixtechnologies.com
- Engineered the Inventory Management System (Django, React, MySQL) with REST APIs and schema migrations, handling inventory valuation across Raw Materials, WIP, and Finished Goods for multi-country operations.
- Created a BOM calculation algorithm in Python for FG and WIP items, processing 2M+ records and reducing turnaround time and manual effort by 70%.
- Developed the Item Generation App integrated with Oracle ERP for item lifecycle management, deployed across 4 countries (China, Thailand, Singapore, Japan) with role-based access control and secured API endpoints.
- Built TracerBee (Django, React), a part traceability application providing end-to-end tracking from Raw Material to Finished Goods, delivered through Agile sprints with peer code review and Git-based workflows.
My Skills
AI & LLM Engineering
Back-end & Databases
Front-end
Cloud, DevOps & Tools
Achievements
Sole Delivery of 5+ AI Systems
Recognised as the only developer to independently deliver five or more production-grade AI and full-stack applications within the first year at InteleANTS.
Promoted to Senior Software Engineer
Awarded for independently managing a cross-border ERP integration project under tight deadlines with zero production incidents.
‘Customer Delight’ ×2
Received twice for delivering web applications with zero escalations from the client.
Projects
A selection of the AI systems, open-source work, and enterprise applications I have built. Source code for the open-source projects is on GitHub.
Advanced RAG Chatbot
GENAI · RAG
Production-grade retrieval-augmented generation service. Combines hybrid BM25 and vector search with cross-encoder re-ranking and context compression, returns cited answers, and measures itself with RAGAS evaluation metrics.
LLM Document Intelligence
GENAI · DOCUMENT AI
Extraction and analysis engine for PDF, DOCX, Excel, CSV, and image inputs. Pulls structured records out of invoices, resumes, and contracts using LLM function calling with Pydantic schema validation.
MCP Server Suite
GENAI · TOOLING
Three Model Context Protocol servers built on FastMCP: SQL Explorer for natural-language database querying, Document Analyzer, and Azure Git Explorer. Includes a chatbot orchestrator driven by function calling, letting non-technical users reach data in plain language.
Call Audit QA Platform
FULL STACK · ENTERPRISE
Full-stack quality-assurance platform with 25+ features, including SSO and JWT authentication, dynamic scoring templates, and real-time analytics dashboards. Adopted by 30+ QA auditors. Extended with TensorFlow models for call sentiment classification and agent performance prediction.
AI Customer Support Chatbot
GENAI · RAG
Multi-channel support assistant with RAG-based knowledge retrieval and automated handoff to a human agent, integrated into Chatwoot through webhooks. Backed by prompt chains that reach 95%+ accuracy on structured extraction across 100+ vendor document formats.
YouTube Video Automation
AUTOMATION · OPEN SOURCE
Script-to-video pipeline that turns a JSON episode script and scene images into a narrated 1080p video with burned-in subtitles, Ken Burns motion, background music, and a generated thumbnail. Supports nine Indian languages and needs no paid APIs.
TracerBee
SUPPLY CHAIN · ENTERPRISE
Part-traceability platform enabling end-to-end tracking from raw material to finished goods in a single click, giving the organisation continuous supply-chain visibility.
Item Generation App
ERP INTEGRATION · ENTERPRISE
Oracle ERP-integrated application managing the full item lifecycle — creation, activation, and updates — with approval routing and BOM validation. Deployed across four countries: China, Thailand, Singapore, and Japan.
Inventory Management
ERP INTEGRATION · ENTERPRISE
Inventory valuation across Raw Materials, Work in Progress, and Finished Goods. Explodes the BOM to summarise purchase cost of buy items and estimate make-item cost including subcontract charges — processing 2M+ records and cutting manual effort by 70%.
More on GitHub
OPEN SOURCE
Further work including a Django REST social networking API with token authentication and request rate limiting, a vendor management system, and a responsive Bootstrap 5 storefront with scroll-reveal animations.