About Skills Experience Projects Education Certifications Contact
🎓 Research Intern @ IIT Roorkee · Open to Work

Mani Ratan

AI/ML Engineer & Data Scientist

Transforming Raw Data Into Intelligent Systems Through Deep Learning, NLP & Production ML Engineering

15+
ML/DL Projects
8.20
CGPA / 10.0
2
Internships
6
Certifications
Who I Am

About Me

I'm a final-year B.Tech student in Computer Science (AI & Data Science) at IIIT Senapati, Manipur. I was awarded a branch upgrade from ECE (VLSI) to CSE (AI-DS) — placing in the Top 5 students — a recognition that reflects my drive and dedication to the field of AI.

Currently, I'm serving as a Research Intern at IIT Roorkee, conducting research on multimodal emotion and personality recognition using physiological signals (EEG, ECG, GSR) from the ASCERTAIN dataset — designing novel hybrid deep learning architectures that surpass published baselines.

I'm passionate about building intelligent systems end-to-end: from data pipelines and model training to deployment and explainability. My expertise spans deep learning, NLP, computer vision, and production ML engineering with a portfolio of 15+ projects including GenAI, RAG, and Agentic AI applications.

"Bridging the gap between research-grade AI and production-ready systems that create measurable real-world impact."
🏆
Top 5 Branch Upgrade
Institutional award — branch upgraded from ECE (VLSI) to CSE (AI-DS) among top 5 students, 2024
🎓
CGPA 8.20 / 10.0
Maintaining strong academic performance with specialized focus on Deep Learning, GenAI and Agentic AI
🔬
IIT Roorkee Research Intern
Multimodal emotion & personality recognition — surpassed baseline F1-score of 0.77 using hybrid architecture
🚀
15+ AI/ML Projects
Portfolio spanning GenAI, RAG, Agentic AI, Computer Vision, NLP and production ML pipelines
Differentiators

What Sets Me Apart

From competitive research internships to real-world ML engineering, here's what makes my profile stand out.

01
🧠
Research-Grade AI

Active Research Intern at IIT Roorkee — building novel multimodal hybrid architectures that beat published baselines in emotion & personality recognition.

02
⚙️
Production ML Engineer

Built end-to-end fraud detection pipelines with >90% Precision & Recall, and edge-optimized AI inference engines within strict hardware constraints.

03
🌐
Full-Stack ML Developer

End-to-end builder: from data pipelines, model training & explainability (SHAP/XAI) to interactive Streamlit dashboards and cloud-native databases.

04
📈
Top Academic Performer

CGPA 8.20 with institutional recognition — awarded branch upgrade to AI-DS among top 5 students for outstanding performance and AI aptitude.

Technical Expertise

Skills & Tech Stack

A comprehensive toolkit spanning machine learning, deep learning, NLP, and production engineering.

🐍
Languages & Databases
Python C / C++ SQL MySQL PostgreSQL OOP
🔧
Frameworks & Tools
TensorFlow PyTorch Scikit-learn Pandas NumPy Streamlit BeautifulSoup Selenium
🤖
Machine Learning
Linear Regression Logistic Regression Decision Trees XGBoost LightGBM PCA / t-SNE Cross-Validation A/B Testing SMOTE / ADASYN
🧬
Deep Learning
ANN CNN (ResNet / AlexNet) RNN / LSTM / GRU Transformers ViT GANs Transfer Learning Fine-tuning TabTransformer
💬
NLP
spaCy NLTK Word2Vec GloVe POS Tagging NER Language Modelling Vectorization
📊
Visualization & XAI
Matplotlib Seaborn Plotly SHAP Git / GitHub Linux (Ubuntu) PyMuPDF
Work History

Experience

Research and industry internships building real-world AI/ML solutions.

Indian Institute of Technology Roorkee
May 2026 – Present · Roorkee, India
Research Intern — Multimodal AI & Physiological Signal Processing
  • Conducting research on multimodal emotion and personality recognition using the ASCERTAIN dataset, analyzing physiological signals (EEG, ECG, GSR, EMO) from 58 subjects across 36 affective movie clips.
  • Designed and implemented a TabTransformer + ResNet hybrid architecture with modality-specific branches and cross-modal feature fusion, surpassing the baseline paper's F1-score (0.77) for personality recognition using 5-Fold Cross Validation.
  • Applied advanced data augmentation techniques including SMOTE, ADASYN and BorderlineSMOTE to address class imbalance in physiological time-series data, improving model generalization across emotion and personality prediction tasks.
Pratinik Infotech
Jan 2026 – Mar 2026 · Remote
Data Science Intern — Fraud Detection & Anomaly Systems
  • Developed an end-to-end Fraud Detection system for credit card transactions, achieving Precision and Recall scores >90% through rigorous model tuning and evaluation.
  • Optimized imbalanced data handling by implementing SMOTE and custom sampling protocols, reducing false negatives in highly skewed financial datasets.
  • Architected a low-latency inference pipeline for near-real-time transaction monitoring, focusing on minimizing financial loss through proactive anomaly detection.
Featured Work

AI/ML Project Portfolio

Intelligent solutions spanning computer vision, NLP, financial ML, and medical AI.

Multimodal Vision Engine
LuminaVision AI — Multimodal Scene Intelligence Engine

Architected a sophisticated multimodal pipeline synchronizing Vision Transformer (ViT) feature extraction with Transformer-based language models for seamless Visual Question Answering (VQA). Engineered a custom cross-domain inference engine optimized for edge deployment within a strict 1GB RAM footprint using manual tensor memory management.

Python PyTorch Transformers ViT Streamlit
View on GitHub
Full-Stack NLP Pipeline
AI-Driven Resume Intelligence & Analytics Engine

End-to-end Information Extraction pipeline using spaCy and NLTK to parse unstructured multi-page PDF resumes with granular skill-gap analytics. Built an interactive Recruiter Analytics Dashboard using Plotly to map candidate distributions and skill densities, backed by a cloud-native TiDB Serverless (MySQL) database.

Python spaCy NLTK TiDB Cloud Plotly PyMuPDF
View on GitHub
Hybrid Ensemble & XAI
Thyroid Sickness Classification Model

Hybrid diagnostic framework integrating CatBoost for gradient boosting and ANN to capture non-linear biomarkers in thyroid pathology detection. Integrated Explainable AI via SHAP to identify TSH and T4 levels as primary clinical predictors — ensuring model transparency and interpretability for medical professionals.

Python PyCaret CatBoost TensorFlow Keras SHAP
View on GitHub
Academic Foundation

Education

2023 – 2027
B.Tech in CSE (AI & Data Science)
IIIT Senapati, Manipur
8.20 CGPA / 10.0
2022
BSEB Class XII — Science
Biseni High School, Sasaram
87% Percentage
2020
CBSE Class X
D.A.V Public School, Sasaram
90.2% Percentage
Credentials

Licenses & Certifications

6
Total Certifications
Demonstrating broad expertise
3
Kaggle Certs
Python, ML, Pandas
3
MongoDB Certs
DB, Connect, Vector Search
2
Platforms
Kaggle & MongoDB
🏅
Kaggle
Python Programming
🏅
Kaggle
Machine Learning
🏅
Kaggle
Pandas
🍃
MongoDB
Database Fundamentals
🍃
MongoDB
Connecting to MongoDB
🍃
MongoDB
AI & Vector Search
Achievements

Notable Highlights

🏆 Institutional

Awarded Branch Upgrade from ECE (VLSI) to CSE (AI-DS) — ranked among Top 5 students at IIIT Senapati, Manipur in 2024, recognizing exceptional AI aptitude and academic performance.

🎓 Academic

Maintaining a CGPA of 8.20 / 10.0 with a specialized focus on Deep Learning, Generative AI, and Agentic AI systems — consistently in the high-performing tier of the cohort.

🚀 Technical

Developed a portfolio of 15+ ML/DL Projects spanning Generative AI, Retrieval-Augmented Generation (RAG), and Agentic AI based applications — demonstrating consistent innovation and execution.

Get In Touch

Let's Connect

I'm actively looking for Data Scientist and AI/ML Engineer roles. Whether you have an exciting opportunity, a research collaboration, or just want to talk AI — I'd love to hear from you.

Currently based in India — open to remote and on-site positions across the globe.

Send Me an Email Let's Connect on LinkedIn ↗