Deepayan
Sarkar.
Data Scientist & Analytics Professional
4+ years turning data into decisions at Accenture → now going deeper into ML & AI at KEDGE. I build pipelines, models, and dashboards that actually ship.
Top 3 Impact Highlights
Records Processed Daily
Processing Time Reduction
Research Time (was 4–8 wks)
Experience
Where I've Built
Things.
- 1Engineered data pipelines processing 10M+ records daily and developed Power BI dashboards tracking 50+ KPIs, reducing processing time by 35% and improving operational efficiency by 20% through ETL optimisation and data visualisation solutions.
- 2Designed and maintained cloud infrastructure on Google Cloud Platform (BigQuery, Dataflow, Cloud Storage) ensuring 99.9% uptime, conducted A/B testing and statistical analysis on datasets with 1M+ rows to optimise product features and improve customer experience by 30%.
- 3Collaborated with 15+ cross-functional stakeholders to deliver 12+ analytical projects, automated reporting processes saving 40+ hours monthly, and mentored junior analysts on data analysis best practices and SQL optimisation techniques.
Impact Highlights
Achievements & Metrics
Results That Speak.
Records processed daily
ETL pipelines at Accenture
ScaleKPIs tracked
Power BI dashboards
ScaleProcessing time reduced
ETL optimisation
ImpactOperational efficiency gain
Data visualisation solutions
ImpactCloud uptime maintained
GCP infrastructure
ReliabilityRows analysed
A/B testing & statistical analysis
ScaleCustomer experience uplift
Product feature optimisation
ImpactStakeholders collaborated
Cross-functional delivery
LeadershipAnalytical projects shipped
Accenture
LeadershipHours saved monthly
Automated reporting
ImpactWeighted F1 Score
L'Oréal multi-label classifier
ML PerformanceProducts classified
L'Oréal hackathon (33 categories)
ML PerformanceRelevance score
AI Persona Bots — BNP Paribas
ML PerformanceCoherence score
AI Persona Bots — BNP Paribas
ML PerformanceFluency score
AI Persona Bots — BNP Paribas
ML PerformanceResearch turnaround
Reduced from 4–8 weeks
ImpactProjects
Selected
Work.
Multi-Label Skincare Product Classifier
L'Oréal Hackathon · KEDGE Business School
- Developed and deployed a multi-label text classification model using LinearSVC and One-vs-Rest classification to classify 6,240 products across 33 categories, achieving a weighted F1 Score of 0.67, in line with industry benchmarks.
- Engineered NLP pipeline with TF-IDF vectorisation (word and character n-grams) and optimised per-class thresholds for improved performance.
Spotify Music Recommendation System
Unsupervised Learning · KEDGE Business School
- Built an unsupervised music recommendation system by applying K-Means clustering to song-level audio features to uncover latent user taste patterns and evaluated performance using the Silhouette, Calinski-Harabasz, and Davies-Bouldin indices.
- Performed feature engineering and preprocessing to enhance clustering stability and designed a similarity-based recommendation approach to enable personalised and cold-start recommendations.
AI Persona Bots for Marketing Research
BNP Paribas & CGI Hackathon · KEDGE Business School
- Built AI-simulated customer personas using Azure AI Foundry and GPT-4o to accelerate credit product launches, processing 2,438 survey responses across 8 distinct customer segments with 88.6% relevance, 97.8% coherence, and 100% fluency scores.
- Engineered end-to-end pipeline with persona generation, NLP-based sentiment analysis, and automated insight synthesis, reducing marketing research time from 4–8 weeks to under 1 hour whilst maintaining high-quality customer simulation accuracy.
China Import/Export Transport Analysis
Tableau Public
- Built an interactive Tableau dashboard exploring China's import/export transport patterns — analysing trade volumes, shipping modes, and commodity flows across global corridors.
- Designed multi-layered filters and drill-down views enabling dynamic exploration of trade data by year, commodity type, and transport mode (sea, air, rail, road), surfacing actionable insights for supply chain analysis.
- Applied calculated fields and LOD expressions to derive year-over-year growth rates and market share breakdowns, visualising shifts in China's top trading partners and strategic export corridors.
Skills
Technical
Arsenal.
Programming & Machine Learning
Data Tools
Cloud & Databases
Core Competencies
Education
Academic
Foundation.
Master of Science in Data Analytics for Business
KEDGE Business School
Master 2nd Year
Bachelor of Technology in Electronics and Communication Engineering
University of Engineering & Management
BTech
Certifications
Google Cloud Certified: Associate Cloud Engineer
Google Cloud
2023
Microsoft Azure AI Fundamentals — AI-900
Microsoft
2026
SQL for Data Science
Coursera · UC Davis
2022
Contact
Let's Build
Something.
I'm actively looking for Data Science internship opportunities from May 2026. Whether you have a role, a project, or just want to talk data — I'd love to hear from you.
Based in Bordeaux, France with France Work Authorisation. Open to internship roles in data science, machine learning, and analytics — in France or internationally.