Welcome, IBM Recruiter — I've highlighted my AI & Enterprise ML projects and skills most relevant to your team.
  █████╗ ██╗  ██╗ █████╗ ███████╗██╗  ██╗    ███╗   ███╗███████╗
 ██╔══██╗██║ ██╔╝██╔══██╗██╔════╝██║  ██║    ████╗ ████║██╔════╝
 ███████║█████╔╝ ███████║███████╗███████║    ██╔████╔██║███████╗
 ██╔══██║██╔═██╗ ██╔══██║╚════██║██╔══██║    ██║╚██╔╝██║╚════██║
 ██║  ██║██║  ██╗██║  ██║███████║██║  ██║    ██║ ╚═╝ ██║███████║
Portfolio OS v2.0 — Type 'help' for commands
akash@portfolio:~$
AI Resume Tailoring Engine
Paste a Job Description below — I'll instantly match it against my skills, highlight relevant projects, and generate a tailored ATS summary.
Skill Match Analysis
Voice Navigation Active
🎤
Listening... say a command
Try saying one of these:
"show projects"Scroll to projects
"show skills"Scroll to skills
"show contact"Contact section
"open terminal"Terminal mode
"best project"Highlights fraud app
"upload job"Opens JD engine
B.Tech Data Science Student & ML Builder

Akash
M S

I build explainable AI systems that solve real-world business problems.
Presidency University, Bangalore · Graduating 2027

I build ML-powered systems that solve real problems — from fraud detection to explainable HR analytics. Certified by Oracle, Google, Meta & AWS. Optum STEM Scholar 2025–26. 4 live deployed apps.

View GitHub LinkedIn
akash@ml-studio — fraud_detection.py
0
GitHub Repositories
0
Certifications
0
ML Projects Built
0
Internships
0
Oracle Certs Active
01About

I'm a Computer Science undergraduate specialising in Data Science, focused on turning raw data into actionable insight and practical solutions. I work with Python, SQL, Pandas, NumPy, Power BI, and ML frameworks.

My experience spans AI-based speech systems, data preprocessing pipelines, analytics workflows, and ML model deployment. I focus on understanding data deeply and solving real business problems — not just collecting certifications.

Actively preparing for Data Analyst, Data Science, and ML roles. Open to internships and entry-level opportunities. Also learning German 🇩🇪 on the side.

📧 ms29akash@gmail.com

Core Tech Stack
PythonSQLPandasNumPy Scikit-learnTensorFlowPower BIStreamlit FastAPIFlaskAWSOracle Cloud MongoDBMySQLRHTML/CSSSHAPNLP
Honors & Education
🏆 Optum STEM Scholar 2025–26
Optum | NASSCOM Foundation · Feb 2026
₹30,000 scholarship — national selection for academic performance and STEM potential.
🌍 4th Innovative Project Expo 2024 — Top 70
Presidency University · Jun 2024
World's Largest Innovative Project Expo finalist.
B.Tech — Data Science
Presidency University, Bangalore · 2023–2027
First Class with Distinction
PU Science (PCMC)
RR Institute of Technology · 2021–2023
Distinction and Honors
02Experience
Jul 2025 – Sep 2025 · 3 months · Bangalore, On-site
Hostingsignal
AI & Data Intern — Speech-to-Text Systems
Internship
  • Situation: Organisation needed multilingual ASR pipelines for Indian language transcription at scale.
  • Processed, cleaned, and annotated large audio datasets to improve transcription accuracy.
  • Evaluated ASR model performance using error analysis and test-set validation across multiple languages.
  • Result: Contributed structured annotation workflows that improved data quality pipeline-wide.
Feb 2025 – Jun 2025 · 5 months · Bengaluru, On-site
CampusX
Logistics & Operations Intern
Internship
  • Situation: CampusX ran 5+ large-scale student programs requiring coordinated logistics.
  • Tracked operational workflows using structured checklists and reporting formats.
  • Collaborated cross-functionally to resolve on-ground issues in real time.
  • Result: Optimised logistics processes reducing delays across all programs.
03Live Deployed Projects
● LIVEML · Finance
Cost-Sensitive Fraud Detection System

Cost-sensitive ML pipeline weighting false negatives far higher than false positives. Real-time dashboard for live transaction scoring with SHAP explainability.

XAIExplainable
LiveReal-time
SHAPAttribution
PythonScikit-learnStreamlitSHAPXAI
● LIVENLP · RecSys
Domain-Agnostic Explainable Recommendation Engine

TF-IDF + semantic similarity engine accepting any dataset. Handles cold-start users and produces human-readable explanations per recommendation.

AnyDomain
ColdStart Ready
XAIPer Item
NLPTF-IDFSemantic SimStreamlit
● LIVEML · HR Analytics
Explainable HR Attrition Risk System

ML classification with SHAP-based explainability identifying at-risk employees. Interactive Streamlit dashboard bridging ML output and business decisions.

SHAPExplainable
HRAnalytics
LiveDashboard
MLSHAPXAIStreamlitHR
● LIVENLP · Safety
Explainable Job Scam Risk Detection System

NLP + ML pipeline classifying job postings as legitimate or fraudulent. Paste any JD — get instant fraud risk score with SHAP feature attribution.

NLPText ML
SHAPAttribution
LivePublic App
NLPClassificationSHAPPython
NLP · ASR · Internship
Speech-to-Text — Indian Languages (Flask)

Flask-based web app using SpeechRecognition API supporting multiple Indian languages with real-time audio processing. Direct production deliverable.

MultiLanguage
FlaskWeb App
RealProduction
FlaskPythonNLPASRHTML/CSS
● LIVENLP · XAI · AgriTech
FarmVoice AI — Explainable Crop Advisory

Explainable AI-powered crop advisory system using NLP + ML that converts natural language into accurate, transparent crop recommendations with SHAP attribution per suggestion.

NLPNatural Lang
XAIExplainable
LiveDeployed
PythonNLPExplainable AIStreamlit
● LIVEML · LLM · Business Intelligence
DecisionIQ — AI Business Intelligence Platform

AI-powered decision intelligence system combining ML + LLM to forecast trends, detect anomalies, and generate actionable business insights. XGBoost + Random Forest + LLM reasoning layer.

LLMInsights
MLForecasting
LiveStreamlit
XGBoostLLMStreamlitRandom Forest
Statistics · Analytics
A/B Testing — Statistical Conversion Analysis

Rigorous A/B testing analysis using statistical hypothesis testing, confidence intervals, and p-value evaluation to drive data-backed business decisions on conversion uplift.

StatsRigorous
CI95%
DataDriven
PythonStatisticsHypothesis TestingPandas
NLP · Chatbot · AgriTech
JeevanMitra AI — Smart Farming Chatbot

Rule-based NLP chatbot serving as a smart farming companion. Provides crop guidance, weather advisory, and farming tips through an intuitive conversational interface.

NLPRule-Based
JSFrontend
SmartAdvisory
JavaScriptRule-based NLPHTML/CSSChatbot
System Architecture
RECRUITER SUMMARY
✔ Python & ML ✔ Explainable AI ✔ NLP ✔ 4 Live Apps ✔ 18+ Certs ✔ Internship XP ✔ Streamlit · FastAPI 🟢 Open to Work
04Project Deep Dive — Fraud Detection System
My most production-realistic ML build — cost-sensitive, explainable, live-deployed
BUSINESS PROBLEM
Banks lose billions to fraud yearly — but standard classifiers optimize for accuracy, not cost.

A model with 99% accuracy that flags no fraud is useless in production. Standard ML treats a missed fraud (false negative) the same as a false alarm (false positive) — but in banking, a missed ₹50,000 fraud costs far more than wrongly flagging a transaction. The real objective isn't accuracy. It's minimizing cost-weighted error.

KEY INSIGHTBuilt a cost-sensitive classifier where FN penalty = 5× FP penalty — mirroring real banking loss ratios.
DATASET & PREPROCESSING
284,807 transactions · 492 fraud cases · 0.17% fraud rate

Extreme class imbalance required deliberate handling — not just oversampling. Applied SMOTE for minority class augmentation combined with class_weight='balanced' in the model. Features were PCA-transformed (V1–V28) plus Amount and Time. Amount required log-scaling; Time required cyclical encoding to capture daily patterns.

1Log-scale Amount (removes right skew)
2Cyclical encode Time (sin/cos of hour)
3SMOTE oversampling on training split only
4StandardScaler — fit on train, transform both
SYSTEM ARCHITECTURE
End-to-end pipeline: raw transaction → fraud verdict + explanation in <20ms
Raw CSV
Input
Feature Eng.
Log · Cyclical
XGBoost
Cost-sensitive
SHAP
TreeExplainer
FastAPI
/predict
Streamlit
Dashboard

Prediction and explanation computed in a single forward pass. SHAP values cached per request. FastAPI handles async inference; Streamlit polls the API every 2s for live demo.

MODEL SELECTION RATIONALE
XGBoost chosen over Random Forest and Logistic Regression — here's exactly why.
ModelRecall (Fraud)LatencySHAP SupportVerdict
Logistic Regression61%2msPartial✗ Too simple
Random Forest79%35ms✓ Full∼ Good baseline
XGBoost ★94%18ms✓ Full✓ Chosen
Neural Network92%210ms✗ None✗ Black box
XGBoost + cost_matrix gave 94% recall on fraud class — every 6 in 100 fraud cases missed vs 39 in 100 with Logistic Regression.
EXPLAINABILITY — WHY SHAP
A fraud alert without a reason is useless. SHAP turns every prediction into a story.

Used TreeExplainer (SHAP) — exact Shapley values for tree models, 10× faster than KernelExplainer. Each prediction outputs: which features pushed toward fraud, by how much, and the baseline fraud rate. This is what regulators and compliance teams actually need.

Example Output — Transaction flagged FRAUD (confidence: 92.3%)
High Amount
+0.41 ↑ fraud
Unknown Merchant
+0.28 ↑ fraud
Late Night (2AM)
+0.19 ↑ fraud
Known Location
−0.09 ↓ fraud
ENGINEERING CHALLENGES & DECISIONS
Every obstacle became an architecture decision. Here's my thinking.
Problem EncounteredDecision MadeWhy This Works
0.17% fraud — model predicts all LegitSMOTE + class_weightSynthetic minority samples balance gradient updates
Equal FP/FN penalty → poor recallCustom cost matrix (FN = 5× FP)Aligns loss function with real business cost
SHAP too slow for real-time (>200ms)TreeExplainer + value cachingExact Shapley in <5ms for tree models
Amount scale dwarfs other featuresLog1p transform + StandardScalerPrevents single feature dominating gradients
Time is circular (23:59 ≈ 00:01)Cyclical encoding (sin/cos)Preserves continuity at day boundaries
RESULTS & METRICS
94% fraud recall · 18ms inference · live on Streamlit Cloud
94%
Fraud Recall
Catches 94 in 100 real fraud cases
92.3%
Avg Confidence
On correctly flagged fraud
18ms
P95 Latency
Predict + SHAP explain
0.87
F1 Score
Fraud class (minority)
920
True Positives
Out of 975 fraud cases
55
Missed Frauds
False negatives (target: <60)
Deployed on Streamlit Cloud · accepts any transaction CSV · full SHAP waterfall chart per prediction · production-realistic pipeline from day one.
04.1Live Demo Gallery
🛡️
Fraud Detection
Upload any CSV · get fraud scores + SHAP explanations per transaction in real time
XGBoostSHAPFastAPI
⬡ Launch App
🔍
SCAMGUARD AI
Paste any job description · instant fraud risk score with NLP feature attribution
NLPscikit-learnSHAP
⬡ Launch App
👥
HR Attrition Risk
Upload HR data · see who will leave + exactly why, ranked by SHAP importance
Random ForestSHAPStreamlit
⬡ Launch App
🎯
Recommendation Engine
Drop any CSV · domain-agnostic recommender with cold-start handling + explanations
TF-IDFNLPStreamlit
⬡ Launch App
04.2ML Workflow — How I Build Every Project
📥
Data Ingestion
CSV · API · DB
🧹
Cleaning
Nulls · Outliers · Types
⚙️
Feature Eng.
Scale · Encode · Select
🤖
Train + Tune
CV · Cost-aware
🧠
SHAP Explain
WHY > WHAT
🚀
Deploy
Streamlit · FastAPI
04.3GitHub Activity
14+ repos · active development
Less
More
05Technical Skills
Machine Learning & AI
Python
92%
Scikit-learn
88%
TensorFlow
75%
SHAP / XAI
85%
NLP / Text ML
80%
Data Engineering
Pandas / NumPy
90%
SQL / MySQL
88%
MongoDB
72%
ETL Pipelines
78%
R Language
68%
Deployment & Cloud
Streamlit
92%
Flask / FastAPI
80%
AWS Cloud
70%
Oracle Cloud OCI
78%
Git / GitHub
85%
Visualisation & BI
Power BI
82%
Plotly / Matplotlib
88%
Seaborn
84%
D3.js
68%
Dashboard Design
80%
06Certifications
Oracle
OCI 2025 Certified Data Science Professional
Oct 2025 · Expires Oct 2027Active
View Badge ↗
Oracle
OCI 2025 Certified Generative AI Professional
Oct 2025 · Expires Oct 2027Active
View Badge ↗
Oracle
OCI 2025 Certified AI Foundations Associate
Oct 2025 · Expires Oct 2027Active
View Badge ↗
Google
Google Data Analytics Professional Certificate
Nov 2025Google
Verify ↗
Meta
GenAI in Data Analytics
Aug 2025Meta
University of London
Machine Learning for All
Sep 2025
Verify ↗
Amazon Web Services
Introduction to IT & AWS Cloud
Nov 2025
Verify ↗
HackerRank
SQLBase Certified Professional
Jul 2025Verified
Verify ↗
SkillUp
Power BI Data Analyst Prep
Sep 2025
Verify ↗
University of Leeds
Programming for Data Science
Aug 2025
Verify ↗
TechXNinjas
Paranox 2.0 — National Innovation Hackathon
Nov 2025National
View ↗
Walmart Global Tech India
Walmart Sparkathon
Jul 2025Hackathon
07Behind the Build — Engineering Wins
Black Box
Explainable
XAI Integration across all 4 projectsAdded SHAP-based explainability to fraud detection, HR attrition, job scam, and recommendation engine — turning model outputs into plain-English business insights recruiters and HR teams can act on.
ML Engineering
Cold Start
Solved
Cold-Start in Recommendation EngineDesigned TF-IDF + semantic similarity fallback for users with zero history. Works across any domain — movies, jobs, products — without retraining or re-deployment.
NLP / RecSys
Siloed Data
Insight
ASR Data Pipeline at HostingsignalStructured annotation workflows for multilingual audio data. Identified recurring transcription error patterns — improved data quality across the full production pipeline.
Data Engineering
Domain Locked
Universal
Domain-Agnostic Pipeline DesignBuilt recommendation and fraud systems that accept any uploaded CSV — not tied to a single dataset. Drag-and-drop ready, zero retraining required.
System Design
Fake Jobs
Detected
NLP Fraud Signal ExtractionEngineered text features capturing linguistic red flags — urgency phrases, missing company details, salary anomalies. Transparent via SHAP attribution for each prediction.
NLP
08Volunteering & Community
🌐
DevSphereIndia
Community Growth & Management
Aug 2025 – Jan 2026 · 6 months
🎓
Internshala
Student Ambassador
Aug 2025 – Present · 9+ months
09Contact
Let's build
something
real.
Open to Data Science, ML, and Analytics roles.
Internships or entry-level. I respond fast.
Quick Facts
LocationBangalore, India
AvailabilityOpen to work ●
Roles SoughtData Analyst / DS / ML
ScholarshipOptum STEM 2025–26 (₹30k)
Oracle Certs3 Active till Oct 2027
DegreeB.Tech Data Science · 2027
LanguagesEnglish · German (Learning)
AI
Akash's AI Assistant
● Powered by Claude · Ask me anything
Hey! I'm Akash's AI assistant — powered by Claude. Ask about projects, skills, certs, or anything about Akash 👋