FloatChat
RAG-over-schema NL→SQL pipeline querying 12,442 ARGO float profiles with a safety layer.
SaanviGrover
Turning data into direction
Machine Learning Engineer, Generative AI Builder, Data Science Practitioner, Python & SQL Developer, RAG + NL→SQL Developer, Problem Solver
I build analytical dashboards, machine-learning systems, and data products that turn complex questions into useful decisions.
- 16
- Projects
- 14
- GitHub Repos
- 9
- ML Projects
- 12
- Live Demos
Featured Projects
Human Skill Extinction Predictor
Forecasts professional-skill demand across one-, two-, and three-year horizons.
HeatFluxAI
City-aware geospatial ML using ERA5, MODIS, and Google Earth Engine.
HR TalentScope
Five-model Flask platform for employee job-change prediction and EDA.
Skills & Expertise
Interview-relevant capabilities grounded in the projects and workflows shown across this portfolio.
Last updated: September 2026
16
Project Case Studies
Complete portfolio case studies
Counted from project pages
9
ML Projects
Regression, classification and NLP
Projects categorized as ML
12
Live Demos
Public interactive deployments
Projects with a live demo link
16
Credentials & Awards
Visible credential and award cards
Matches the credentials section
Technical Skills
12 SkillsNo self-rated percentages
PythonProgramming
SQLProgramming
Pandas & NumPyData Analysis
Data Cleaning & EDAData Analysis
Probability & StatisticsStatistics
Hypothesis TestingStatistics
Regression & ClassificationMachine Learning
Feature EngineeringMachine Learning
Model EvaluationMachine Learning
Model Selection & TuningMachine Learning
Data VisualizationVisualization
BI DashboardsVisualization
ML Interview Coverage
Concrete portfolio evidence across the workflow interviewers usually probe.
Problem framing & EDA
TalentScope | DataStory AI
Feature engineering
HeatFluxAI | Kshetra Sense
RAG & Generative AI
FloatChat | ToneLeaf
Validation strategy
Life Expectancy | HeatFluxAI
Model selection & tuning
Kshetra Sense | Life Expectancy
Deployment & inference
GradMit | ToneLeaf
Skills by Category
Programming
2 SkillsPython, SQL, data querying
Data Analysis
2 SkillsPandas, NumPy, cleaning, EDA
Statistics
2 SkillsProbability, inference, hypothesis tests
Machine Learning
4 SkillsModels, features, evaluation, tuning
Visualization
2 SkillsData storytelling, Tableau, Power BI
Project Portfolio
16 ProjectsA collection of data-driven solutions, ML models, analytics dashboards, and systems I've built.
16
Projects
9
ML Projects
12
Live Demos
14
GitHub Repos
Project Scope
Generative AI / RAG
Machine Learning
Data Visualization
HR Analytics
Climate Analytics
Deployed workforce intelligence platform
ML In progress
Human Skill Extinction Predictor
AI-Powered Future Skill Demand & Obsolescence Forecasting Platform
Forecasts professional-skill demand across one-, two-, and three-year horizons.
95
Canonical Skills
2,179
Skill-Year Rows
3Y
Forecast Horizon
PythonML ForecastingPandasTrend Analytics
Deployed guarded recommendation workspace
ML
Kshetra Sense
Ranked Crop Recommendations from Soil & Climate Data
Guarded crop recommendations across 22 classes with calibrated, abstaining inference.
99.55%
Test Accuracy
22
Crop Classes
4
Models Compared
PythonStreamlitScikit-LearnRandom Forest
Deployed safety-oriented triage interface
Systems
TriageTide
Explainable, Local-First Symptom Triage Prototype
Deterministic symptom-urgency guidance with explainable rules and explicit safety boundaries.
19
Automated Tests
3
Triage Levels
Local
Core Decision Path
PythonStreamlitSupabaseGroq
Held-out and cross-validation evaluation
ML
Life Expectancy ML Capstone
End-to-End Regression, Classification & Explainable Predictions
End-to-end life-expectancy modeling with cross-validation, tuning, and saved artifacts.
0.9923
Test ROC-AUC
0.9898
CV ROC-AUC
2,938
Records
PythonScikit-LearnJupyterGridSearchCV
Built under academic requirements during the Masai AI/ML course.
Published NBA dashboard on Tableau Public
Visualization
Sports Analytics Dashboard
Interactive NBA Performance Analytics Dashboard
Interactive Tableau analysis of NBA players, teams, scoring, and game outcomes.
8+
Visualizations
Player + Team
Analysis Views
Live
Dashboard Filters
TableauData VisualizationDashboard Design
News-source distribution output
Analytics
TrendPulse
Real-Time News Data Pipeline
Live-news API pipeline for cleaning, feature engineering, EDA, and saved visual outputs.
4
Pipeline Stages
Live
API Source
2
Visual Outputs
PythonPandasRequestsMatplotlib
Built under academic requirements during the Masai AI/ML course.
What Drives My Work
Turning complex data into impact
Building systems that scale
Curiosity + code = solutions
Detailed Project Briefs
Every project card now links to a full project description with metrics, highlights, and technologies.
// Keep learning. Keep building.
while(alive) {
code();
learn();
build();
}
Project Insights
Data-driven insights from my projects and work.
16
Project Case Studies
Detailed portfolio projects
Counted from project pages
9
ML Projects
Regression, classification and NLP
Projects categorized as ML
12
Live Demos
Public interactive deployments
Projects with a demo link
14
GitHub Repositories
Linked source repositories
Projects with source code
12
Advanced Case Studies
Technically deep project pages
Projects marked Advanced
16
Credentials & Awards
Visible credential and award cards
Counted from displayed credentials
Selected Technical Proof
Six projects with concrete validation, scale or deployment evidence
HeatFluxAI
25 cities and 26 engineered features
Benchmarked up to 8 regressors with city-aware GroupKFold across ~4,500 real observations.
Life Expectancy Capstone
0.9923 test ROC-AUC
Tuned a reusable Random Forest pipeline with five-fold validation on 2,938 WHO records.
Kshetra Sense
99.55% test accuracy across 22 crops
Compared 4 tuned classifiers and shipped ranked top-five recommendations.
ToneLeaf
745/745 deterministic audit cases
Built explainable, local-first NLP with separate sentiment, threat and distress signals.
HR TalentScope
19,158 records, 5 models, 10 EDA modules
Connected workforce analysis, feature engineering, validation and exportable predictions.
HelixForge
75,145 graph edges and 10 core algorithms
Implemented a C++17 genome-assembly pipeline with De Bruijn graphs and N50 analysis.
Engineering Capabilities
What I can build, with projects that prove each capability
Evaluation-first machine learning
Leakage-aware splits, cross-validation, GridSearchCV and held-out metrics across regression and classification.
Evidence: HeatFluxAI | Life Expectancy | Kshetra Sense
Production-facing data products
Turned models and analysis pipelines into usable Streamlit, Flask, Next.js and FastAPI workflows.
Evidence: TalentScope | EDA Studio | ToneLeaf
Explainable and responsible AI
Added feature importance, influential language cues, confidence context and explicit usage limitations.
Evidence: ToneLeaf | Kshetra Sense | HeatFluxAI
Data engineering and analytics
Built API ingestion, cleaning, feature engineering, interactive dashboards and exportable analysis.
Evidence: TrendPulse | DataStory AI | Sports Analytics
Portfolio Composition
Where the 16 case studies concentrate technically
Machine Learning
9 projects
Analytics
4 projects
Visualization
1 project
Systems & Algorithms
2 projects
Built for applied DS/ML roles
81% of the portfolio is focused on machine learning or analytics. The emphasis is on validation, explainability, deployment and decision support—not notebook-only demos.
ML Evaluation, in Context
Comparable only within each task—never blended into a fake score
HeatFluxAI
City-aware validation
Up to 8 regressors | GroupKFold | 25 cities
Life Expectancy Capstone
0.9923 test ROC-AUC
0.9898 mean CV ROC-AUC | 2,938 WHO records
Kshetra Sense
99.55% test accuracy
99.49% mean CV accuracy | 4 tuned classifiers
GradMit
91.2% test accuracy
0.975 ROC-AUC | 500 records
HR TalentScope
5-model comparison
Cross-validation | 19,158 training records
ToneLeaf
745/745 audit checks
81/81 labelled cases | explainable local inference
Accuracy, ROC-AUC and validation scope measure different things and are intentionally not ranked against each other.
Stories Worth Discussing
Four strong interview narratives and why the technical choices matter
HeatFluxAI
Leakage-aware geospatial regression
Grouped cities during validation across ~4,500 observations so test locations never appear in training.
HR TalentScope
From 19,158 records to probability outputs
Combined 10 EDA modules, engineered features, 5 models and downloadable predictions.
Life Expectancy Capstone
A four-stage ML workflow
Moved from cleaning and baselines to ensembles, tuning, saved artifacts and guarded LLM explanations.
ToneLeaf
NLP with explicit safety boundaries
Separates sentiment, threats and distress cues; stores no text and documents where the model can fail.
How I Work
Engineering habits visible across the portfolio
Validate before deploying
ValidationUsed GroupKFold, stratified splits, five-fold CV and held-out tests instead of reporting training accuracy.
Keep pipelines reproducible
ReproducibilityPersisted preprocessing and model artifacts so training and inference apply the same transformations.
Make results inspectable
ExplainabilityUsed feature importance, influential words, model comparisons and metric context to explain outputs.
Finish with a usable product
Product thinkingProjects end in a prediction, ranking, dashboard, export or API rather than stopping at a notebook.
From Student to Data Analyst
A timeline of how I evolved from a curious student into a data analytics and machine learning practitioner through coursework, certifications, competitions, mentorship, and industry experience.
2024
Academic Foundation
Built a strong foundation in Mathematics, Statistics, and Logical Reasoning, scoring 92.2% in PCM.
92.2%
in PCM
July 2024
University, JIIT
Pursuing Integrated M.Tech in Computer Science while developing expertise in Data Analytics, Machine Learning, and Software Development.
Duration
July 2024 – May 2029
(Int M.Tech)
Current CGPA
8.3
Dec 2025
Analytics Competition Achievement
Secured Rank 1 in Round 1 of Statistella BASH 8.0 and advanced as a finalist in a national-level data analytics and statistics competition.
Statistella BASH 8.0
National Data Analytics Competition
Feb – Aug 2026
Vishlesan I-Hub Foundation, IIT Patna
AI & ML Certification (IIT Patna)
Vishlesan I-Hub Foundation, IIT PatnaCompleted the Certification in AI & Machine Learning (Hindi) under Vishlesan i-Hub, IIT Patna / Masai.
2 Feb 2026 – 23 Aug 2026
9.16
CGPA
3
Certificates of Excellence
28 May - 8 Jul 2026
View Mentorship Certificate
Persistent Systems Mentorship
Worked with industry mentors at Persistent Systems, gaining hands-on exposure to software engineering practices, project development, and professional workflows.
Project: SkillLens / HSEP
28 May 2026 - 8 July 2026
7 - 20 Jul 2026
AI Data Analytics Internship
Completed a remote AI Data Analytics internship, contributing assigned analytical work while demonstrating reliability, communication, responsibility, and teamwork.
Aug 2026
ANALYTICA 2026 Finalist
Selected as a finalist at ANALYTICA 2026, a national-level data science hackathon. Built FloatChat — a natural-language querying system for ocean data using RAG + NL→SQL.
Team Entropy · PS26-033
National-Level Data Science Hackathon
12,442
Float Profiles
36/36
Safety Tests
Learning. Building. Deploying. Improving.
Certifications & Achievements
Continuous learning, verified proof, and real impact.
ANALYTICA
2026 Finalist
ANALYTICA 2026 — National Finalist
ANALYTICA 2026 · National-Level Data Science Hackathon
Selected as a finalist at ANALYTICA 2026 for FloatChat — a natural-language querying system for ocean data. Built the AI backend: RAG-over-schema NL→SQL pipeline with a multi-layered query safety system. Team Entropy (PS26-033).
RAGNL→SQLLLMPostgreSQLQuery Safety
Credential value
TypeNational hackathon finalist
EvidenceANALYTICA 2026 competition result
National-level hackathon finalist
View CertificatePersistent Systems Mentorship
Persistent Systems Ltd.
Completed the Persistent Systems mentorship from 28 May 2026 - 8 July 2026 while developing the Human Skill Extinction Predictor (HSEP) under industry mentorship.
Data AnalyticsMachine LearningIndustry Mentorship
Credential value
TypeIndustry mentorship
EvidenceMentorship completion certificate
Completion documented
View Mentorship CertificateGoogle
Google Advanced Data Analytics
Google / Coursera
Learned data cleaning, analysis, visualization, statistics, and machine learning using Python, Tableau, and real-world datasets.
PythonStatisticsMLTableau
Credential value
TypeProfessional learning credential
EvidencePublic verification link
Verified Certificate
View CertificateIIT Patna × MasaiAI & ML Evaluation Series
Vishlesan i-Hub, IIT Patna / Masai
Three Certificates of Excellence from the Certification in AI & Machine Learning (Hindi), Feb – Aug 2026. CGPA 9.16. Presented together as one learning journey.
02 May 2026
Evaluation 1
Program evaluation credential
Verify Eval 131 July 2026
Evaluation 2
Program evaluation credential
Verify Eval 201 July 2026
Evaluation 3
Program evaluation credential
Verify Eval 3Artificial IntelligenceMachine LearningCGPA 9.16
Verified Certificate
3 verified milestonesInAmigos
Foundation
AI Data Analytics Internship Portfolio
InAmigos Foundation
A complete record of the remote AI Data Analytics internship from 7 July 2026 - 20 July 2026: completion, contribution recognition, and a professional recommendation from the Founder and CEO.
Document 1
Internship Completion
AI Data Analytics internship
View CertificateDocument 2
Certificate of Appreciation
Contribution and commitment
View RecognitionDocument 3
Letter of Recommendation
Professional recommendation
View LetterAI Data AnalyticsProfessionalismTeamworkContribution
Three documents available
3 verified milestones
IIT BHU
Statistella BASH 8.0
Business Associates Special Hours / IIT BHU
Secured Rank 1 in Round 1 and advanced as a finalist in a national-level data analytics and statistics competition.
Data AnalyticsStatisticsCompetition
Credential value
TypeCompetition achievement
EvidenceLeaderboard and participation certificate
Verified Certificate
View CertificateHackerRank
SQL Certification Series
HackerRank
A three-level HackerRank SQL credential series covering core querying, joins, subqueries, aggregations, analytical logic, and advanced database problem-solving.
Document 1
SQL (Basic)
Core querying and relational concepts
View BasicDocument 2
SQL (Intermediate)
Joins, grouping, and subqueries
View IntermediateDocument 3
SQL (Advanced)
Complex analytical query logic
View AdvancedSQLJoinsSubqueriesAdvanced Queries
Verified Certificate
3 verified milestonesHackerRank
Python (Basic)
HackerRank
Verified Python fundamentals including data types, control flow, functions, collections, and problem-solving with clean Python syntax.
PythonProblem SolvingCore Syntax
Credential value
TypeTechnical skills assessment
EvidencePublic HackerRank certificate
Verified Certificate
View Certificate
LiveMasai Live – AI Job Builder
Masai Live
Completed the Masai Live AI Job Builder program, building and deploying AI-powered tools for job applications, resume generation, and automated outreach workflows.
Consistent Performer
Problem Solver
TrailblazerAI WorkflowsPrompt EngineeringAutomation
Verified Certificate
View CertificateGoogle
Google AI Essentials
Google / Coursera
Gained foundational knowledge of AI concepts, prompt engineering, AI ethics, and responsible AI practices.
AI FundamentalsPrompt EngineeringAI Ethics
Credential value
TypeProfessional learning credential
EvidencePublic verification link
Verified Certificate
View CertificateYou keep learning, you keep growing.
Every certificate adds one more layer of practical skill, stronger judgment, and confidence with real data problems.
16
Credentials
8
Issuers
30
Skill Tags
2
Achievement Years
More credentials coming soon.
New projects, certificates, and analytics wins will keep joining this collection.
Get In Touch
Have an internship, research opportunity, collaboration, or project in mind? Send me a message directly.
Open to opportunities
Your message goes straight to my inbox.
I'm interested in AI/ML, data science, analytics, research, and software development opportunities.
IndiaJIIT studentUsually replies in less than 24 hours
To ______
From ______
Saanvi's inbox
Ready when you are
Send me a message
Tell me about the role, research opportunity, or project you have in mind.
Professional enquiries only. Message delivery is handled by FormSubmit, so share only the contact details needed for a reply.