MS Data Science · ASU · Graduating May 2026

Aniket
Deshpande.

Data Scientist  ·  ML Engineer  ·  Applied AI & LLM Practitioner

Hi! I'm Aniket, a data scientist who builds production ML systems that survive the real world, not just notebooks: drift-aware fraud detectors, low-latency LLM microservices, and RAG pipelines shipped, monitored, and battle-tested. I'm currently finishing my Master's in Data Science, Analytics, and Engineering at Arizona State University (graduating May 2026).

If you're hiring engineers who are adaptive, quick to learn, curious and comfortable building from scratch, iterating fast, and enjoying the journey from experiment → production and keep them healthy at scale, let's chat. I don't stop at a working model, I try to make sure it's reliable, measurable, and built to last.

View My Work Resume PDF
95%RUL Accuracy
83%Fraud Recall
3.8GPA @ ASU
5+Live Projects
Aniket Deshpande
AWS ML Specialty
AWS Certified ML Specialty

Selected Work

Projects that ship &
actually scale.

End-to-end systems from raw data to deployed APIs. Each is live and production-tested.

LLM NLP App

01 — Featured

LLM-Powered NLP Microservice

Production FastAPI + Gradio stack delivering text summarization & sentiment via transformer models. Blue-green deploys via GitHub Actions with fallback routing for 100% uptime.

FastAPIGradioTransformersCI/CDDocker
  • 100% service availability via fallback routing between local models and OpenAI/OpenRouter APIs.
  • GitHub Actions CI/CD handling 500+ daily API requests at sub-second latency.
  • Modular UI for sentiment analysis, summarization, and text rewriting.
Stockio

02

Stockio — Forecast & Sentiment

Multi-ticker dashboard with Prophet/ARIMA/XGBoost forecasts, VADER sentiment, and 93% anomaly detection precision. SHAP explainability for transparent insights.

ProphetXGBoostStreamlitSHAP
  • 93% anomaly detection precision via ensemble forecasting with guardrails.
  • Cache-aware pipeline; deployed via Azure DevOps to Hugging Face Spaces.
  • SHAP explainability for transparent, risk-aware stakeholder visualizations.
RAG QA

03

RAG-QA over SEC Filings

CPU-only RAG using MiniLM + FAISS. Automates EDGAR ingestion for Risk & MD&A sections, returning citation-grounded extractive answers.

FAISSMiniLMEDGARRAG
  • Automated pipeline ingesting Risk Factor & MD&A sections from SEC 10-K filings.
  • Every answer includes a citation-grounded source passage for verifiability.
  • BLEU/ROUGE evaluation. Fully CPU-based — no GPU required.
Fraud Detection

04

Drift-Aware Fraud Detection

Recall-optimized pipeline on 280K+ transactions. PSI drift tracking with adaptive threshold recalibration restoring ~25% recall on drifting streams.

XGBoostPSI DriftSMOTEScikit-learn
  • 83% recall at 0.17% fraud rate using prequential time-ordered evaluation.
  • Adaptive recalibration restores ~25% recall; keeps manual review under <1%.
  • SMOTE vs class-weight analysis: class weights superior for PCA features.
Speech Emotion Recognition

05

Speech Emotion Recognition

Audio → MFCC maps → CNN/RNN classifier. Real-time prediction via mic or .wav with confidence scores per emotion class. Deployed on Hugging Face Spaces.

LibrosaMFCCPyTorchGradio
  • Hybrid CNN-RNN on MFCC + mel-spectrogram features via Librosa.
  • Real-time prediction with confidence scores per emotion class.
  • Confusion matrix error analysis for robust multi-class debugging.

Capabilities

Full-stack ML from
data to deployment.

End-to-end expertise across the modern ML stack modeling, infrastructure, and shipping.

⚙️

ML & Modeling

PyTorchTensorFlowScikit-learnXGBoostLightGBMLSTMCNN / RNNTime-seriesFeature Eng.
🧠

LLMs & NLP

TransformersHugging FaceLangChainRAG / FAISSBERT / GPT-2LLaMA-2 / T5Prompt Eng.Sentiment Analysis
☁️

MLOps & Cloud

AWS SageMakerS3 / LambdaEMR / GlueDockerGitHub ActionsAzure DevOpsTerraformCI/CD
📊

Data & Analytics

PythonR / SQLApache SparkPandasEDAA/B TestingPSI DriftCalibration
🖥️

APIs & Web

FastAPIFlaskGradioStreamlitREST APIsHTML / CSS / JS
📈

Visualization & BI

PlotlyMatplotlibTableauPower BISHAPSeaborn

Work History

Where I've made
a real impact.

ML deployment experience across EdTech and renewable energy real systems, real outcomes.

Aug 2024 – Present · Scottsdale, AZ

Instructional Design Assistant

EdPlus @ Arizona State University

  • Analyzed 10K+ learner interaction events from adaptive learning & gamified modules across 15+ production courses using Python & SQL.
  • Engineered 20+ behavioral & psychometric features for ML-based early-risk detection, improving QA readiness by ~30%.
  • Deployed models on AWS SageMaker; ran A/B experiments improving outcomes across 60+ courses.
  • Authored end-to-end experiment reports and documentation reducing onboarding time for new collaborators.

May 2021 – Aug 2021 · Pune, India

Data Analyst Intern

Suzlon Energy Pvt. Ltd.

  • Built Python RUL predictive models on SCADA & CMS time-series data 95% failure detection accuracy.
  • Automated anomaly detection pipelines in SQL; built Tableau dashboards reducing manual diagnostics by 25%.
  • Preprocessed high-volume sensor data for multi-turbine time-series modeling.

Academic Background

Built on a strong
foundation.

Progressive education from CS fundamentals all the way to large-scale ML systems.

M.S. · Current

Data Science, Analytics & Engineering

Arizona State University · Tempe, AZ

GPA: 3.8 / 4.02024 – May 2026

ML Systems · Statistical Modeling · Big Data · Cloud Computing · Data Mining · Design of Experiments · Data Visualization

PG Diploma

Artificial Intelligence & Machine Learning

MIT World Peace University · Pune, India

GPA: 3.7 / 4.02022 – 2023

Deep Learning · NLP · Transformers · Speech Emotion Recognition research capstone

B.C.A.

Bachelor of Computer Applications

MIT World Peace University · Pune, India

GPA: 3.8 / 4.02019 – 2022

Programming · DSA · Databases · Software Engineering · Internship @ Suzlon Energy

🏅

AWS Machine Learning Specialty (ML-SC01)

Amazon Web Services · Verified

🎓

High-Dimensional Data Analysis

HarvardX

💼

Data Science Professional Certificate

IBM

🔬

Applied NLP / Cloud Computing

Infosys Springboard

About Me

More than
the résumé.

I'm Aniket, a data scientist finishing my MS at ASU in May 2026, focused on building ML that doesn't just score well on benchmarks but runs in production and drives real decisions.

My path: started with a BCA in Computer Science, where I built my foundations. The real spark came at Suzlon Energy, working on turbine health modeling that’s when I saw how data could directly impact real-world systems. I went on to pursue a PG Diploma focused on deep learning and NLP, and now at ASU, I’m building and shipping ML pipelines while running A/B experiments that influence real student outcomes at scale.

Outside the terminal: swimming, trekking, photography, making tech explainer videos. Open to full-time Data Scientist, ML Engineer, and Applied AI roles, graduating May 2026.

Let's connect.

Graduating May 2026. Actively seeking full-time Data Scientist, ML Engineer, and Applied AI roles. Or just use the chat assistant below it knows everything about me!

✉️
Emailadeshp61@asu.edu
→ 💼
LinkedInAniket Deshpande
→ 💻
GitHubView Code & Repos
→ 📄
ResumeDownload PDF
→

Quick Tour

This portfolio is hand-built with HTML, JavaScript, and CSS, you'll find a curated set of projects, many deployed and live.

They sit on free-tier hosts, so a minute or two to wake is normal.

My resume is available for download on the site, and the About page lists relevant coursework and GPAs.

🤖
Ask About Aniket AI Assistant
🤖
Hi! 👋 I'm an AI assistant with complete knowledge of Aniket's background, skills, projects, and experience.

Ask me anything — what he's built, his availability, or how to reach him!

Responses based on Aniket's verified information