Sudhin S.
Researcher, AI/ML @ TCS Research (TRDDC)
Architecting Agentic AI & Human-Centric Systems.
I'm a Researcher at TCS Research (TRDDC), where I work on data-driven decision models, knowledge graphs, and AI for legal use cases. My path here ran through an M.Tech in Data Science at IIT Palakkad and a B.Tech in CSE (AI) at Amrita — I specialize in LLM-based orchestration, retrieval evaluation, and turning abstract quality goals into concrete, trackable metrics.
/ Career Log
Date: Jul 2026 — Present
Researcher
TCS Research (TRDDC)
Working on decision models, knowledge graphs, and AI for legal use cases.
Date: Jul 2025 — May 2026
Placement Coordinator, Data Science M.Tech
Career Development Centre, IIT Palakkad
Coordinated placement activities and industry outreach for the Data Science M.Tech cohort.
Date: Jul 2025 — May 2026
Deputy PG Secretary
Student Affairs Council, IIT Palakkad
Supported postgraduate student affairs, representation, and event coordination.
Date: Jul 2024 — Jul 2026
M.Tech, Data Science
IIT Palakkad · CGPA 7.44
Research in LLM-based agents, RAG evaluation, and legal-domain NLP. Also served as Placement Coordinator & Deputy PG Secretary.
Date: Sep 2022 — May 2023
Competitive Programming Team Lead
ACM Student Chapter, Amritapuri
Led the ACM competitive programming team and mentored juniors through contests and problem-solving sessions.
Date: 2020 — 2024
B.Tech, CSE (Artificial Intelligence)
Amrita Vishwa Vidyapeetham · CGPA 8.24
Led the ACM Student Chapter competitive programming team (40+ members) and ran algorithm workshops for 100+ students.
/ Research Focus
Decision Models
Building data-driven models that turn unstructured signals into actionable, trackable decisions.
Knowledge Graphs
Modeling complex entity relationships for efficient traversal, search, and structured reasoning.
AI for Legal
Evaluation pipelines for legal-domain RAG — measuring faithfulness and statute-level retrieval quality across large case corpora.
/ Major Projects
Legal RAG Evaluation Framework
Modular evaluation pipeline over 50K+ Indian legal documents, benchmarking retrieval robustness, faithfulness, and statute overlap across Llama and DeepSeek.
Llama / DeepSeek / FAISSAgentic AI for Finance
Multi-agent system ingesting unstructured news and social data into investment insights via knowledge graphs — built for Inter IIT Tech Meet '25.
LangChain / AutoGenDeepfake Audio XAI
Deep learning pipelines (CNNs, RNNs) on in-the-wild datasets, applying LIME for transparent, debuggable model explanations.
PyTorch / LibrosaAI Career Guidance
End-to-end web app using a KNN-based prediction engine, deploying RESTful APIs for real-time inference to 100+ active users.
Django / scikit-learn/ Publications
Leveraging Stock Prediction Using News Reports
End-to-end ML pipeline combining NLP-based sentiment analysis with time-series models for stock movement prediction.
ADMM: Computational Analysis of LASSO and RIDGE on Fraud Detection
LASSO and Ridge regression with ADMM optimization for fraud detection, with parallelized computation.
Initiate Connection
Whether you want to discuss AI architecture, explore service design methodologies, or collaborate on rigorous research, I am always open to connecting.