Scroll for stats

Tanmay Kumar Sahu

Tabla rhythms become spectrograms, then MIDI.

Documents become grounded answers.

Faces become real-time emotion signals.

The story starts where disciplines collide.

0

GATE Qualified

DA & CS (2026)

0.00

CGPA

B.Tech IT

0+

ML Projects

Deployed

0+

Attendee Event

Led

0 Years

Hindustani

Classical Training

0+

Design Assets

Created

About

Where classical music meets machine learning.

I'm a final-year IT student at GGV who combines a six-year background in Hindustani classical vocals with deep learning engineering. I built a CNN+BiLSTM pipeline that transcribes tabla audio into DAW-compatible MIDI because I believe the most interesting problems live at the intersection of disciplines. GATE qualified in both Data Analytics and Computer Science. Proficient in PyTorch, LangChain, FastAPI, and RAG architectures.

Name

Tanmay Kumar Sahu

Degree

B.Tech in Information Technology

University

Guru Ghasidas Vishwavidyalaya, Bilaspur

CGPA

7.81

Focus

AI / ML / Deep Learning / RAG

Location

Bilaspur, Chhattisgarh, India

Selected Work

Projects as systems, not just repositories.

Audio ML

TablatoDraw / Audio-to-MIDI Transcription

Teaching machines to read Indian classical percussion.

CNN + BiLSTM pipeline classifying 6+ tabla stroke patterns from log-mel spectrograms with PCEN normalization.

Onset detection via Librosa for time-aligned MIDI generation.

Deployed as a FastAPI REST endpoint for real-time DAW integration.

PyTorchLibrosaFastAPIAudio MLBiLSTM

RAG Systems

Enterprise Document Intelligence Chatbot

RAG pipeline for grounded enterprise Q&A.

LangChain + ChromaDB vector store for semantic search over PDF knowledge bases.

Gemini API with prompt-engineered templates for citation-backed responses.

Streamlit interface with document upload and conversational Q&A.

LangChainChromaDBGemini APIStreamlitRAG

Computer Vision

Emotion Recognition AI

Real-time facial emotion classification at 7-class precision.

CNN with transfer learning trained on 35,000+ images.

15% robustness improvement via data augmentation and Dropout.

Live Streamlit interface for webcam and image input.

TensorFlowOpenCVTransfer LearningStreamlit

Skills

A practical stack for building intelligent products.

Languages & Tools

PythonSQLCFastAPIStreamlitGitJupyter

Frameworks & Libraries

PyTorchTensorFlowScikit-learnOpenCVLibrosaLangChainHugging FaceGemini APIPandasNumPyFAISSChromaDB

Competencies

RAGVector DatabasesPrompt EngineeringData AnalysisMachine LearningFoundation ModelsMulti-agent Systems

Education & Journey

A timeline of sound, systems, and self-study.

2019

Hindustani Classical Sangeet Diploma, multiple Distinctions

2021

Class 12 PCM, 88.6%, CBSE

2022

Joined B.Tech IT at GGV

2025 Feb

Joined UDAAN University Magazine as Graphic Designer

2025 Mar

Cultural Coordinator, EQUILIBRIO Techfest (2,000+ attendees)

2025 Jul-Dec

Built Emotion Recognition AI

2026 Jan-Apr

Built Audio-to-MIDI Transcription system (flagship project)

2026 Mar

GATE Qualified, DA and CS both

2026 Apr

Built Enterprise Document Chatbot

Achievements

Proof points from exams, cloud AI, and classical training.

Trophy

GATE DA & CS 2026

Top national exam in data science and computer science. Qualified in both streams through self-study.

Cloud

AWS AI Practitioner

GenAI, LLMs, RAG, Amazon Bedrock, Responsible AI.

Music

Hindustani Classical Diploma

Six years, multiple distinctions.

Leadership & Beyond Code

Visual storytelling, performance, and large-team execution.

Graphic Designer, GGV

UDAAN Magazine

Created 30+ digital assets using Adobe Suite and Canva.

Cultural Coordinator

EQUILIBRIO Techfest

Directed 50+ member team for 10+ events, 2,000+ attendees.

Performing ArtsGraphic DesignClassical Music

Contact

Let's build something meaningful.

Open to Data Science and ML roles, research internships, and AI/ML R&D opportunities.

2026 Tanmay Kumar Sahu. Built with Next.js, exported for GitHub Pages.