SABYASACHI SHADANGI

Principal AI Architect

Cloud AI Platforms | ML & Deep Learning | Generative AI & LLMs | Agentic AI | MCP | A2A | Advanced RAG | MLOps & Distributed Training | Architectural Design

📧 sabyasarangi7@gmail.com 📱 +91-9284031539 🎓 M.Tech (BITS Pilani) 🏢 Coforge 📚 Pursuing PhD, Data Science
Sabyasachi Shadangi

Executive Summary

Principal AI Architect with 20+ years of enterprise technology leadership, including 10+ years in Data Science / ML and 4+ years architecting Generative & Agentic AI systems at production scale. Proven track record of designing cloud-agnostic AI platforms (AWS, Azure, GCP) that deliver measurable business outcomes — 98.5% fraud-detection accuracy, 95%+ RAG retrieval precision, 40% relevance lift in agentic chatbots, 92%+ forecasting accuracy.

Trusted advisor to Fortune 500 clients across BFSI, Healthcare, Telecom, Retail, FMCG, and Logistics, with deep expertise in RFP solutioning, cost optimization, and translating business strategy into scalable technical architecture.

Profile Snapshot

Current Role

Principal AI Architect — GenAI Practice, Coforge (Encora Inc.) since Nov 2024

Experience

20 years total | 4 years GenAI Architect | 10 years Data Science

Domains

BFSI, Wealth Management, Telecom, Healthcare, e-Commerce, FMCG/Retail, Logistics, Government

Roles Held

Enterprise Architect, Principal AI Architect, Lead Data Scientist, Associate Architect, Program Manager

Recent Project Highlights

MultiTenant Agentic AI Based Service

Healthcare, HR, e-Commerce — FastAPI + LangGraph + MCP powered by GPT-4o; 40% relevance improvement, 30% hallucination reduction.

Multi-Modal RAG Platform

Cross-cloud (AWS, Azure, GCP) retrieval with LangChain + Pinecone/Weaviate/Chroma; 95%+ retrieval accuracy.

Real-Time Fraud Detection

PyTorch + XGBoost ensemble on 1.5 TB transactions at 98.5% accuracy, integrated with NICE Actimize for AML.

AI Demand Forecasting

Hybrid stack (TFT, DeepAR, TimesFM 2.0, Moirai 2.0, Prophet, LightGBM, SARIMAX) — 92%+ accuracy at SKU level.

Micro-Diff

AI system detecting micro-level semantic and structural differences between document versions.

AI-Based Global Translator

Real-time browser data translation across 30 global languages.

Core Capabilities

Generative AI & LLMs

Claude, GPT, Llama, Gemini, Mistral

Agentic AI MCP & A2A

Autonomous AI systems with reduced hallucinations

Advanced RAG Architectures

Multi-modal search & retrieval (95%+ accuracy)

ML & Deep Learning

Fraud detection, ensemble models (98%+ accuracy)

Cloud AI Platforms

AWS Bedrock, Azure ML, Vertex AI

AI Tech Stack

Python, FastAPI, LangChain, LangGraph, LlamaIndex, Pinecone, Chroma, Weaviate, OpenSearch, Azure AI Search

Frameworks & Tools

Orchestration

LangGraph, LangChain, LlamaIndex, LiteLLM, CrewAI, AutoGen

Vector DBs

Pinecone, Qdrant, Chroma, Weaviate, FAISS, OpenSearch, Azure AI Search, Vertex AI Vector Search

LLMs

GPT, Claude, Llama, Gemini, Mistral

Prompt Engineering

Chain-of-Thought, Few-shot, Structured Output, Function Calling, ReAct

Fine-tuning

LoRA / QLoRA, OpenAI fine-tuning API, Azure ML

Cloud

AWS Bedrock, SageMaker, Azure OpenAI, Vertex AI

Automation

n8n, Make, Zapier, FastAPI custom pipelines

Forecasting Models

LightGBM, Prophet, Neural Prophet, TFT, DeepAR, TimesFM 2.0, Moirai 2.0, Gemini 1.5 Pro, SARIMAX, ARIMA

Professional Experience

Coforge (Encora Inc.)
Principal AI Architect — GenAI Practice
Nov 2024 – Present
Multi-domain
  • Lead GenAI architecture across BFSI, Healthcare, and Retail Fortune 500 clients.
  • Implemented live use cases for summarization, chatbots, and multi-modal content extraction using LLM + Advanced RAG on GCP, Azure, AWS.
  • Own RFP solutioning, cost estimation, and resource planning for multi-million-dollar AI engagements.
  • Engaged on Agentic AI, MCP, A2A ,RAG, Hybrid Search, LangChain, LangGraph, AWS Bedrock/Kendra, Vector DBs, LlamaIndex.
  • Validation matrices: ROUGE, BLEU, METEOR; observability via LangSmith.
Tech Mahindra
Principal AI Architect
Previous
Logistics, Telecom, BFS, Auto, Government
  • Designed and shipped an agentic AI chatbot using FastAPI + LangGraph + MCP, integrating GPT-4o — 40% relevance gain, 30% hallucination drop.
  • Architected cloud-based payment-gateway infrastructure for BFS and Wealth Management.
  • Integrated NICE Actimize for real-time fraud prevention, AML, trading surveillance, and sanctions monitoring.
  • Led monolithic mainframe → microservices cloud migration; Docker, Kubernetes (EKS/AKS/GKE).
  • Pre-sales, auditing, RFP estimation, and development team guidance.
EPAM
AI Architect / Data Scientist
Previous
BFS, Wealth Management
  • Built Advanced RAG pipelines with LangChain, vector DBs (Chroma, Pinecone, Weaviate, FAISS), and FastAPI inference endpoints across AWS / Azure / GCP.
  • CNN networks for image classification/detection; combined with metadata for multi-modal classification.
  • Context-based summarization with LSTM networks; sentiment-score comparison.
  • Designed auto-scaling groups, AWS RDS multi-AZ HA, CloudFront edge delivery.
Accenture
AI Architect / Data Scientist
Previous
BFS, e-Commerce, Healthcare, FMCG
  • Feature engineering on 1.5 TB of credit-card transactions; fraud-detection models with XGBoost + Deep Learning.
  • Customer-churn analytics in banking and e-commerce.
  • Sentiment analysis using deep learning with Keras.
  • Strong Python + async/await fluency; debugged FastAPI services in production.
TCS
Lead Data Scientist / AI Lead
Previous
BFS, Telecom, Retail, Media
  • Object classification, image reconstruction with autoencoders, and binary segmentation.
  • Facial detection/classification/recognition/verification using deep learning.
  • Visual Question Answering using LSTM + deep learning.
  • Twitter network analysis and Genetic Algorithm for TSP.
Infosys
Associate Architect
Previous
Telecom BSS / OSS
  • Architected COTS / MOTS / NOTS solutions; TMForum Open APIs and TMF ODA.
  • Established SoE / SoR / SoI frameworks; data governance for quality, security, and compliance.
  • Managed 4 development teams; pre-sales, estimation, design, client advisory.
IBM
Senior Developer
Previous
Logistics, Telecom, Insurance
  • SOLID and 12-Factor principles; security, reliability, resilience features.
  • Java + WebSphere Portal; contributed to digital-transformation initiatives.
Asian CERC (Religare Technova)
Developer
Previous
Trading, Wealth Management
  • Built the Trade Anywhere trading application with Java, JDBC, JSP, Servlets, REST, XML, JSON, JMS.

Personal & Fun Projects

  • Text Summarization — LSTM networks with context-based sentiment preservation.
  • Visual Question Answering — Deep learning with LSTM.
  • AI Video Content Generator — Auto-creates themed videos from source clips with matching background music.
  • Twitter Network Analysis — Centrality measures & community detection.
  • Genetic Algorithm for TSP — Traveling Salesperson optimization.

Education

PhD in Data Science — Pursuing
M.Tech, Data Science — BITS Pilani
B.E., Computer Science & Engineering — BPUT, Odisha

Certifications

NVIDIA — Generative AI & LLMs TOGAF 9.2 Azure Solutions Architect Expert (AZ-305) Azure Data Scientist Associate (DP-100) Google Professional Data Engineer PMP® PRINCE2 CSM CSPO CSP SAFe CAL-1 Accenture Technical Architect Accenture Data Architect IBM Certified Solution Developer S.E.C.U.R.E. Your Future — INSEAD Executive Programme — IIM Calcutta

Delivery Methodology

  • High-quality, client-appreciated deliverables across 20+ Fortune 500 engagements.
  • End-to-end ownership of the AI solution lifecycle — ideation → PoC → production → MLOps.
  • Proven ability to translate business requirements into scalable, cost-optimized architectures.
  • Deep expertise in FinOps for AI — token economics, inference cost tuning, GPU optimization.
  • Excellent stakeholder management; CXO and engineering audiences alike.