My WorkMes réalisations

Projects & BuildsProjets & Réalisations

Enterprise systems from BoA & Cognizant, plus open-source ML/AI/Data projects on GitHub.Systèmes entreprise BoA & Cognizant, et projets ML/IA/Data open-source sur GitHub.

Case Study · Banking

Workflow automation for 100K+ internal users

ProblemManual task allocation, slow reporting, and sensitive access rules across banking operations.

BuildASP.NET Core APIs, SQL Server, OAuth 2.0/JWT/RBAC, Splunk dashboards, Azure DevOps and Jenkins deployment pipeline.

Impact40+ secured endpoints, 500K+ daily transactions, 30-40% less manual reporting, and 40% faster release cycles.

Case Study · Data Engineering

Enterprise ETL and reporting pipelines

ProblemLarge reporting workloads and multi-system data movement for banking, logistics, and construction platforms.

BuildSSIS/SSRS, SQL Server stored procedures, Web API/WCF integrations, batch processing, and monitoring workflows.

ImpactMillions of records processed daily, reporting latency reduced by 2 business days, and BI runtime improved by 35%.

Case Study · AI / Big Data

Academic AI systems with production-style patterns

ProblemTurn ESILV coursework into practical, recruiter-readable systems with clear data flow and measurable evaluation.

BuildKafka, PySpark, Hadoop/HDFS, Power BI, BERT, Scikit-learn, MLflow, SHAP, TensorFlow/Keras, and PyTorch notebooks.

ImpactPublic GitHub evidence for data pipelines, NLP, MLOps, deep learning, and computer vision work.

Open-source evidence Public repositories are linked where code can be shared. Enterprise projects stay confidential, but include architecture, responsibilities, and impact.

GitHub Profile → GNN Project → Big Data → MLOps →

Professional & Enterprise Projects

ENTERPRISE Bank of America · Feb 2022 – Sep 2025

Banking Workflow Automation Platform

Enterprise task allocation and workflow engine for internal banking operations — 100,000+ users, secure RBAC, and CI/CD-driven deployments.

Tech Stack

Responsibilities

Impact

ASP.NET CoreC#SQL ServerEF CoreAzure DevOpsJenkinsOAuth 2.0 / JWTCQRSIISSplunk

Designed scalable RESTful APIs for secure integration across enterprise banking systems
Led workflow automation platform development supporting 100,000+ users
Implemented OAuth 2.0 / JWT with RBAC for sensitive data access control
Optimized SQL Server queries — improved response time by 40%
Managed CI/CD with Azure DevOps and Jenkins; handled IIS and load balancing

100K+ users40+ API endpoints−30–40% manual processing500K+ txns/day

ENTERPRISE Cognizant · Jul 2019 – Feb 2022

Enterprise Logistics & Financial Backend

Scalable backend services for logistics and financial platforms — high-volume batch processing, data ingestion, and multi-system REST API integration.

Tech Stack

Responsibilities

Impact

ASP.NETC#SQL ServerSSISREST APIsBatch ProcessingCI/CDData Ingestion

Built backend services for enterprise logistics and financial systems
Developed data ingestion and batch processing workflows for high-volume transactions
Integrated external systems via REST APIs with retry logic and error handling
Supported CI/CD pipelines and managed production deployments

3M+ records/dayMulti-system integrationReduced processing time

ENTERPRISE LTI Mindtree · Jul 2017 – Jul 2019

Construction Project Tracking & Analytics

End-to-end enterprise system for construction project monitoring — SSIS-driven ETL, workflow tracking modules, and SSRS dashboards for management KPIs.

Tech Stack

Responsibilities

Impact

SQL ServerSSISSSRSETL PipelinesASP.NET / C#Database ArchitectureWorkflow Engine

Designed database architecture and reporting schema using SQL Server
Built workflow tracking modules for task assignment and team monitoring
Implemented SSIS ETL processes for multi-source data integration
Generated SSRS reports and dashboards for management KPI tracking

Automated reportingMulti-source ETLReal-time dashboards

Academic & Open Source — GitHub Projects

M.Sc. Data Science & AI — ESILV Paris (2025–2026) & personal open-source work

GNN · GITHUB srinivasand04/tacticai-lite · May 2026

TacticAI-Lite — Football GNN Tactical Analysis

Graph Neural Network system that models football matches as player graphs and predicts tactical outcomes. Implements GCN and GAT architectures on StatsBomb open data, achieving ~38-44% accuracy — 4× above the 9% random baseline across 9 formation classes.

Tech Stack

Responsibilities

Impact

PyTorch GeometricGCNGATPythonStatsBombmplsoccerScikit-learnJupyter Notebook

Built player-graph construction pipeline from StatsBomb open event data using mplsoccer
Implemented GCN (Kipf & Welling 2017) and GAT (Veličković et al. 2018) architectures with PyTorch Geometric
Trained multi-class formation classifier across 9 tactical formations (4-3-3, 4-4-2, 3-5-2, etc.)
Evaluated with accuracy, confusion matrix, and per-class F1 — GAT outperformed GCN by ~4–5%
Visualised pitch graphs and attention weights using mplsoccer and matplotlib

GAT ~38–44% accuracy4× over random baseline9 formation classesStatsBomb open dataGNN on sports analytics

DATA ENG · GITHUB srinivasand04 · Updated Jan 2026

SQL to NoSQL Migration Framework

Intelligent schema transformation framework — workload analysis, automated cost estimation, and multi-cloud comparison (MongoDB, DynamoDB, CosmosDB) for migration decisions.

Tech Stack

Responsibilities

Impact

PythonSQL ServerMongoDBDynamoDBCosmosDBSchema AnalysisCost EstimationMulti-cloud

Built intelligent SQL schema parser and NoSQL transformation engine
Implemented workload analysis to recommend optimal NoSQL target
Cost estimation module comparing MongoDB Atlas, DynamoDB, and CosmosDB
Multi-cloud comparison report generation with migration recommendations

Automated schema transformMulti-cloud comparisonCost estimation

NLP · GITHUB srinivasand04/NLP_Project · Updated Dec 2025

Yelp Review Sentiment Analysis — Transformers

NLP project for sentiment classification on Yelp reviews using Transformer models (BERT / DistilBERT) with HuggingFace fine-tuning and evaluation on large-scale review data.

Tech Stack

Responsibilities

Impact

BERTDistilBERTHuggingFaceTransformersPythonPandasJupyter NotebookFine-tuning

Fine-tuned BERT and DistilBERT on Yelp review dataset for sentiment classification
Preprocessing pipeline: tokenization, padding, attention masks
Evaluated models with accuracy, F1-score, and confusion matrix
Compared Transformer-based vs traditional ML baselines

BERT fine-tunedYelp datasetTransformer NLP

MLOPS · GITHUB srinivasand04 · Updated Feb 2026

Employee Attrition Prediction — MLOps

End-to-end MLOps pipeline to predict employee attrition using machine learning — feature engineering, model training, experiment tracking with MLflow, and automated deployment.

Tech Stack

Responsibilities

Impact

PythonScikit-learnXGBoostPandasNumPyMLflowSHAPMLOps Pipeline

Built end-to-end MLOps pipeline for employee attrition classification
Feature engineering and preprocessing with Pandas / NumPy
Trained Random Forest and XGBoost models with hyperparameter tuning
Tracked experiments, metrics and model artifacts using MLflow
SHAP-based model interpretability and feature importance analysis

MLflow trackingSHAP explainabilityEnd-to-end MLOps

DEEP LEARNING · GITHUB srinivasand04 · Updated Jan 2026

Stock Market ML & Deep Learning Prediction

Advanced stock market prediction using multiple deep learning architectures — LSTM, GRU, CNN, and Transformer models with comparative analysis on historical price data.

Tech Stack

Responsibilities

Impact

LSTMGRUCNNTransformerPythonTensorFlow / KerasPandasJupyter Notebook

Implemented LSTM, GRU, CNN and Transformer models for time-series stock prediction
Data preprocessing: normalization, sliding window, train/test split
Comparative model evaluation — RMSE, MAE, directional accuracy
Visualized predictions vs actual prices with Matplotlib

4 DL architecturesTime-series forecastingComparative analysis

BIG DATA · GITHUB srinivasand04/BigData_Project · Updated Feb 2026

AQI Big Data Pipeline — Kafka & PySpark

Real-time air quality index (AQI) big data pipeline — Kafka streaming ingestion, PySpark processing on Hadoop/HDFS, and Power BI dashboards for environmental analytics.

Tech Stack

Responsibilities

Impact

Apache KafkaPySparkHadoop / HDFSPower BIDAXPythonShell ScriptData Lake

Built Kafka streaming pipeline for real-time AQI sensor data ingestion
Processed and transformed data using PySpark on Hadoop cluster
Stored processed data in HDFS data lake for batch analytics
Built Power BI dashboard with DAX measures for AQI trend visualisation

Real-time streamingHadoop/HDFSAQI dashboardESILV project

COMPUTER VISION · GITHUB srinivasand04/ComputerVision · Updated Jan 2026

Computer Vision — Data Labs

Computer vision experiments and data lab notebooks — image classification, object detection, and CNN model exploration using OpenCV and PyTorch as part of ESILV coursework.

Tech Stack

Responsibilities

Impact

OpenCVPyTorchPythonScikit-learnCNNImage ClassificationJupyter Notebook

Image preprocessing and augmentation pipelines with OpenCV
CNN architecture design and training for image classification tasks
Model evaluation: accuracy, confusion matrix, precision/recall
Experimented with transfer learning using pretrained PyTorch models

CNN classificationTransfer learningESILV coursework

View All on GitHub → Tout voir sur GitHub →