DEV CARD NO. 001
Swarnajit Roy

HELLO, I'M

SWARNAJIT ROY

AI / ML Engineer & Systems Developer

I turn machine learning research into systems people can use. Specializing in retrieval-augmented generation (RAG), computer vision ensembles, and automated analytics pipelines.

RAG SYSTEMS COMPUTER VISION APPLIED ML
08 FEATURED BUILDS
04 INTERNSHIPS
02 PUBLICATIONS
Explore Featured Builds Download Résumé
Bengaluru, India
Christ Univ · CGPA 3.55 / 4.00
NO. 001

About Me

SWARNAJITDEX
Swarnajit Roy
CGPA
3.55/4.00
BUILDS
8

Developer — AI & ML Type

ZONE: BENGALURU   ABILITY: RAG & COMPUTER VISION

I’m an Electronics & Computer Science Engineering student (B.Tech ECS - AI & ML) at Christ (Deemed to be University), Bengaluru. My focus is on turning machine learning research into reliable software that people can inspect, evaluate, and use in practice.

I’ve worked across RAG document processing pipelines (Minu Travels), automated EDA & machine-learning workflows (UST), multi-backbone deepfake detection ensembles (IEEE Computer Society), and quantum steganography research (SaaAI '25).

Beyond coding, I serve as Wing Head for Skill Development & Training at the Industry Institute Interaction Cell (IIIC), leading technical workshops for students.

  • Bengaluru, India
  • B.Tech ECS (AI & ML)
  • 4 AI / ML Internships
  • English, Hindi, Bengali
SELECTED SYSTEMS

Featured Builds & Projects

Detailed builds across RAG, computer vision, predictive ML, and analytics automation. Click any system to inspect technical architecture.

FLAGSHIP 2025
DEVELOPER TOOLS

Tekshila

An AI documentation and code-quality assistant that turns repository analysis into review-ready pull-request feedback on GitHub.

LLM APIs GitHub API Code Intelligence
Porygon-Z Delivery surface: Annotated PRs inside GitHub workflow
FLAGSHIP · UST APR - JUL 2026
ANALYTICS PIPELINE

AI Auto-EDA & Modelling

An internal analytics workflow at UST Global that moves raw datasets from profiling to explainable baseline models in one Streamlit interface.

Python Streamlit Scikit-learn Groq API
Rotom Impact: Reduced initial dataset profiling from days to hours
FLAGSHIP · IEEE CS APR - JUL 2025
COMPUTER VISION

Deepfake Detection System

A deep learning ensemble combining EfficientNet, ResNet, and Vision Transformers (ViT) for real-versus-synthetic media classification.

PyTorch EfficientNet ResNet ViT
Metagross Architecture: 3-Model ensemble with weighted sampling
LAB PROJECT 2025
LLM & RAG

Agragrati

A resume-intelligence prototype for automated skill extraction, structured relevance scoring, and external job discovery via APIs.

Groq API Python Streamlit Docker
Smeargle Features: Skill extraction & inspectable matching score
RESEARCH PAPER DEC 2025
PREDICTIVE ML

Battery Life Estimation

Comparative study of single and stacked regression models using a Battery Degradation Factor feature formulation. Presented at SMAMIA 2025.

XGBoost Scikit-learn Ensemble
Elekid Benchmark: Stacked R² ≈ 0.98 | MAE ≈ 1.5 hours
LAB PROJECT 2025
LLM & RAG

Mentora

A document-learning assistant powered by IBM Watsonx.ai that turns educational PDFs into grounded active-recall summaries and quizzes.

IBM Watsonx.ai Python Streamlit
Slowking Pipeline: PDF Extraction -> Grounded Chunks -> Quizzes
LAB PROJECT 2025
NLP & MULTIMODAL

Multimodal Summarizer

A unified summarization interface (Poke-Summarizer) for documents, audio, video, and live news using Groq LLM inference.

Groq API Media Pipelines Multilingual
Chatot Modalities: Documents, Audio, Video, Live News
SIGNAL TOOLKIT 2025
ANALYTICS & SIGNALS

Planetary Seismic Detection

An interactive signal-processing toolkit using ObsPy to analyze STA/LTA trigger events across Mars and Moon seismic datasets.

ObsPy Signal Processing Streamlit
Deoxys Domains: Mars & Moon STA/LTA waveform triggers
MOVESET

Skills & Capabilities

FIRE

Machine Learning

PyTorch
—
Scikit-Learn
—
Hugging Face
—
WATER

LLM & RAG

RAG Systems & Vector Search
—
FAISS Embeddings
—
Watsonx.ai & Groq APIs
—
ELECTRIC

Languages

Python
—
SQL
—
Java / C++
—
STEEL

Delivery & Tools

Streamlit & FastAPI
—
Docker & Git
—
OpenCV & ObsPy
—
PSYCHIC

Capabilities

  • Production RAG System Architecture
  • Computer Vision Deepfake Ensembles
  • Automated EDA & Baseline Modeling
  • Signal Processing & Waveform Triggers
DRAGON

Expertise

  • Data Structures & Algorithms
  • Quantum Hybrid Steganography
  • Feature Engineering & Stacking
  • Technical Training & Workshop Leadership
FAIRY

Soft Skills

  • Technical Workshop Leadership
  • Research Communication
  • Cross-Functional Handoff
  • Mentoring
TRAINER PATH

Experience & Education

DEGREE PROGRAM

Christ (Deemed to be University), Bengaluru

Bachelor of Technology in Electronics & Computer Science Engineering (AI & ML)

PERIOD 2023 - 2027
CGPA 3.55 / 4.0
Charmander NO.001

IEEE CS Bangalore

AI Research Intern

Developed deepfake detection ensembles combining EfficientNet, ResNet, & Vision Transformers with weighted sampling.

PyTorch Vision Transformers OpenCV
Apr 2025 – Jul 2025
KARNATAKA
Charmeleon NO.002

AICTE - Edunet

AI & ML Intern

Architected end-to-end regression ML prediction workflows, data preprocessing pipelines, and MAE evaluation interfaces.

Regression Scikit-learn Pandas
Jun 2025 – Jul 2025
REMOTE
Charizard NO.003

Minu Travels

AI Chatbot Developer

Designed & shipped a production RAG system processing 100+ business documents with FAISS vector retrieval & confidence fallbacks.

RAG FAISS LLMs Python
Oct 2025 – Nov 2025
WEST BENGAL
Mega Charizard X NO.004

UST

AI / ML Intern

Automated exploratory analysis & baseline ML workflows into a reusable Streamlit tool, reducing initial dataset review time from days to hours.

Python Streamlit Scikit-learn Groq API
Apr 2026 – Jul 2026
WEST BENGAL
EVIDENCE & RECOGNITION

Research, Awards & Gallery

Swarnajit Roy presenting research at SaaAI '25
RESEARCH PRESENTATION

SaaAI '25 International Conference

Swarnajit Roy at Productathon AI, IIT Roorkee
PRODUCT BUILD

Productathon AI, IIT Roorkee

Swarnajit Roy receiving university award
UNIVERSITY RECOGNITION

Christ University Award Ceremony

Research Publications

PRESENTED · AUG 2025

Quantum Hybrid Steganography Model

SaaAI '25 (International Conference on Sustainable Analytics & AI)

Proposed a quantum-classical hybrid steganography framework using simulated quantum key distribution.

Scizor88.06% Security PorygonSSIM ~ 0.996
CONFERENCE PAPER · DEC 2025

ML Models for Battery Life Estimation

SMAMIA '25 (Smart Materials & Industrial Applications)

Compared single & stacked regressor models using a domain Battery Degradation Factor feature formulation.

DragoniteR² ≈ 0.98 DialgaMAE ≈ 1.5 Hours

Recognitions & Awards

  • WINNER

    Gen'AI Work-a-thon

    Christ University

  • FIRST PLACE

    Extra-Curricular Activities Award

    CUESTIC EC Club, Christ University (2024-2025)

Credentials & Leadership

  • AI Agents Bootcamp (LetsUpgrade / NSDC / GDG)
  • Generative AI Foundations (AWS Academy)
  • Build RAG Chatbot with Python (LetsUpgrade)
  • Tata Group Data Analytics Job Simulation (Forage)
LEADERSHIP ROLE

Wing Head — Skill Development & Training

Industry Institute Interaction Cell (IIIC), Christ University

LIVE SYNC LOGS

GitHub Contribution Energy Matrix

Real-time commit telemetry & 52-week activity logs directly from GitHub repositories.

@coder-royswarnajit

Trainer Activity & Live Telemetry

VISITS: --
GH VIEWS: --
STAR REPO --
YEAR COMMITS --
CURRENT STREAK --
LONGEST STREAK --
PEAK DAY LOG --

52-Week Energy Matrix Grid

[INITIALIZING POKÉDEX TELEMETRY LINK...]
LESS MORE

Language Stack Telemetry

[SYNCING LANGUAGE TELEMETRY...]

Live Recent Commit Stream

[FETCHING RECENT COMMITS FROM GITHUB API...]
EXPLORE FULL GITHUB PROFILE
DEV CENTER

Get in Touch

Interested in AI/ML engineering or applied research roles? Reach out directly or grab my resume below.

royswarnajit341@gmail.com
+91 62893 02303
Download Résumé (PDF)