Open to full-time opportunities
AI/ML Engineer · Data Scientist · Mechanical Engineer · Researcher
Building AI that knows
when to be certain.
I design reliable AI systems that combine multimodal learning, retrieval, and confidence-aware decision making—grounded in an interdisciplinary foundation spanning data science and engineering.
01 / Perspective
My work sits at the intersection of machine intelligence and engineering judgment. I build models that do more than predict—they measure uncertainty, retrieve evidence, and know when to step back.
Making confidence
measurable.
Research on trustworthy multimodal systems, evidence retrieval, calibration, and selective decision making.
01
Published paper · JENRS 2026
Reliability-Aware Multimodal AI
Retrieval-Augmented Reliability-Aware Selective Inference for Visual Classification
A framework that retrieves external visual evidence, estimates instance-level reliability, and decides whether to accept, answer with caution, or abstain when evidence is weak.
88.88% accepted accuracy at 89.04% coverage21.48% relative reduction in accepted wrong answers
02
Published paper · IEEE SIEDS 2026
From Weak Embeddings to Trustworthy Multimodality
Modality-Agnostic Learning with Confidence-Aware Fallback
A confidence-aware framework combining multimodal representation learning, uncertainty estimation, calibration, and fallback validation for more trustworthy decisions.
Published in IEEE XploreDecision-Making Track · SIEDS 2026
03 / Project portfolio
Ideas made
practical.
Projects across multimodal AI, computer vision, applied machine learning, analytics, recommendation, and intelligent agents.
View all on GitHub ↗01NeuralSeek Ignite · Agentic AI
Equipment Failure Intelligence Agent
A reliability-aware predictive-maintenance agent that combines calibrated failure-risk models, explainability, input-support checks, maintenance knowledge, and human-review escalation.
Calibrated RF: 0.857 F1 · 0.904 PR-AUC
02Computer Vision · Deep Learning
SmartLeafNet
Hybrid plant-disease classification using EfficientNetB3 and ResNet50 feature fusion, PCA, SMOTE, and Optuna-tuned XGBoost.
98% accuracy · 0.95 macro-F1
03Multimodal Learning · Embeddings
UniBind
Unified multimodal learning with contrastive embeddings, similarity-based retrieval, and LLM-supported reasoning across modalities.
Retrieval + confidence-aware decisioning
04Applied ML · Molecular Data
Shear Viscosity Prediction
Regression analysis for alkane shear viscosity using thermophysical features, statistical exploration, and ensemble machine-learning models.
Random Forest RMSE 0.0493
05Predictive Analytics · Classification
Customer Churn Analytics
A statistically validated churn-classification workflow that transforms customer signals into actionable retention insights.
Test error approximately 1.21%
06Recommender Systems · Applied AI
Product Recommendation
A recommendation-system project focused on matching product choices with user preferences and behavioral data.
Personalized, data-driven discovery
04 / Mechanical Engineering & MBSE
Where systems thinking
became second nature.
My mechanical engineering background shapes how I approach AI: define the system, understand constraints, validate every interface, and design for real-world operation.
Featured engineering project · 01
Eco-Friendly E-Bike Taxi
A multidisciplinary vehicle-development project connecting requirements, mechanical design, system architecture, analysis, and sustainable mobility into one traceable engineering concept.
Recognized by Capgemini Engineering for the effective application of MBSE to an eco-friendly urban mobility solution.
View MBSE project on GitHub ↗Model-Based Systems EngineeringSystem ArchitectureRequirements & TraceabilityMechanical Design
CAD & product design
Autodesk · SolidWorks · CATIA V5 · PTC Creo · Product development
Simulation & analysis
ANSYS · Mechanical systems · Engineering analysis · Verification & validation
MBSE
CATIA MagicGrid · Requirements analysis · Functional decomposition · Traceability · System architecture
Systems thinking
Interface definition · Trade-off analysis · System reliability · Sustainability-oriented design · Technical documentation
05 / Experience
Applied AI.
Research depth.
Summer 2026
NeuralSeek AI · Ignite Internship Program
AI & Data Science Intern
Developed a predictive-maintenance intelligence agent that combined calibrated Random Forest and XGBoost models, SHAP explanations, risk banding, evidence retrieval, and human-review escalation to convert equipment data into actionable inspection recommendations.
View project on GitHub ↗Jan—May 2026
University of Massachusetts Dartmouth
Graduate Research Assistant
Built confidence-calibrated multimodal pipelines, embedding-based evidence retrieval, and offline evaluation safeguards for reliable AI decision making.
Mar—May 2023
Dr. Mahalingam College of Engineering and Technology
Learning & Development Associate — Data & Analytics
Created Python analytical pipelines and automated reporting workflows, translating financial and operational data into stakeholder decisions.
06 / AI & data toolkit
How I
build.
LLM & agentic AI
LangChain · RAG · LLM agents · Tool use · Prompt engineering · Structured outputs
ML & deep learning
PyTorch · TensorFlow · Hugging Face · XGBoost · Model evaluation · Calibration
Data & engineering
Python · SQL · Pandas · NumPy · PySpark · REST APIs · Docker · Git
Analytics
Statistical validation · Hypothesis testing · Tableau · Power BI · Excel
07 / Education
M.S. Data Science
University of Massachusetts Dartmouth · GPA 3.95
Post Graduate Program, Data Science
VIT Bangalore · GPA 3.68
B.E. Mechanical Engineering
Dr. Mahalingam College of Engineering and Technology · GPA 3.56
Selected certifications
Generative AI for Project Managers · PMIProject Management · IIT KanpurModel-Based Systems Engineering · SUNY Buffalo08 / Contact
Have a hard problem
worth solving?
I'm exploring full-time opportunities in AI/ML engineering, data science, applied AI research, and interdisciplinary engineering.