Satya Siba Nayak — Portfolio

SHEET 01

Profile & specifications

Field	Value
Program	BCA — Computer Science
Institution	MIT-WPU, Pune
Year	Junior year
Location	Pune, India
Focus	Open source · browser extensions
Status	Building & maintaining

I'm Satya — an undergraduate BCA (Computer Science) student at MIT-WPU, Pune, currently in my junior year.

Most of what I build ends up open source: small, local-first tools that are honest about what they do and don't ask for your data. Browser extensions over platforms, plain code over frameworks I don't need.

Day to day that means maintaining real repositories — reviewing contributor pull requests, triaging issues, and shipping features people actually install and use.

SHEET 02

Bill of components

Parts list for the stack I build with most. (Placeholder — swap in your real list.)

No.	Component	Type	Notes
01	JavaScript	Language	Primary
02	HTML / CSS	Markup / Style	Core
03	Browser Extension APIs	Platform	Chrome · Firefox · Edge
04	Git & GitHub	Tooling	Version control
05	Supabase	Backend	Auth / DB
06	REST / CDN APIs	Integration	Live data
07	CS Fundamentals	Coursework	DS&A, DBMS
08	Titanic EDA →	Coursework	Missing-value imputation
DATASET train.csv · 891 rows · 12 columns · PassengerId, Survived, Pclass, Name, Sex, Age, SibSp, Parch, Ticket, Fare, Cabin, Embarked CONTEXT Lab 2 · IMLP practical INSPECTION head / tail, shape, column names, info, describe, dtypes; numerical vs categorical columns MISSING VALUES Age 177 · Cabin 687 · Embarked 2 METHODS Drop incomplete rows, mean fill and median fill (Age), mode fill (Embarked) — each applied on a separate copy of the data TARGET Survived (target) vs the remaining 11 columns as features RESULT Dropping every incomplete row leaves only 183 of 891 records; the fill methods keep all 891. STACK pandas Open notebook →
09	Iris EDA →	Coursework	Pandas basics · Iris data
DATASET Iris.csv · 150 rows · 6 columns · Id, SepalLengthCm, SepalWidthCm, PetalLengthCm, PetalWidthCm, Species FORMAT Practical 1 — ten questions (Q.1–Q.10), each answered with a short Pandas snippet COVERS head, random sample, shape, column names, dtypes, missing values, unique species and counts, mean / median / std, dropping a column RESULT No missing values · 3 species × 50 rows · SepalWidthCm mean 3.05, median 3.0, std 0.43 STACK pandas Open notebook →
10	Binary Classification →	Coursework	Logistic regression · toy data
DATA Small toy set — one feature with values 1–10; label 0 for 1–5 and 1 for 6–10 SPLIT 80 / 20 · 8 train · 2 test · random_state 42 MODEL LogisticRegression (scikit-learn), scored with accuracy RESULT Accuracy 1.0 on the 2 test samples — a tiny set, so it shows the workflow rather than real-world performance DEMO Predicts for a new student with value 7: class 1, probability ≈ 0.85 STACK NumPy · scikit-learn Open notebook →
11	Titanic Logistic Regression →	Coursework	Survival prediction
DATASET Titanic-Dataset.csv · target: Survived FEATURES Pclass, Sex, Age, SibSp, Parch, Fare PREP Missing Age and Fare filled with the median · Sex encoded male → 0, female → 1 SPLIT 80 / 20 · random_state 42 · 179 test passengers MODEL LogisticRegression (max_iter 1000) · accuracy and classification report RESULT Test accuracy 0.81 · precision / recall: died 0.81 / 0.88, survived 0.80 / 0.72 DEMO 1st class, female, age 28, travelling alone, fare 100 → survived, probability ≈ 0.94 STACK pandas · NumPy · scikit-learn Open notebook →
12	Insurance Charge Prediction →	Coursework	BMI → charges · regression
DATASET insurance.csv · 1,338 rows · age, sex, bmi, children, smoker, region, charges INPUT → TARGET BMI (single feature) → insurance charges SPLIT 80 / 20 · 1,070 train · 268 test · random_state 42 MODELS Linear regression, then polynomial regression at degrees 2, 3, 4 and 5 (PolynomialFeatures + LinearRegression) METRICS R², MAE, MSE, RMSE — train vs test R² compared per degree RESULT Linear test R² ≈ 0.04; polynomial test R² 0.01–0.03. Higher degrees didn't help, so BMI alone explains little of the variation in charges. DEMO Predicts the charge for a new BMI of 30: ≈ 13,100–13,600 across the four degrees STACK pandas · NumPy · Matplotlib · scikit-learn Open notebook →

SHEET 03

Deployed systems

FIG. 1 — FLUX (BROWSER EXTENSION) · FLAGSHIP

Zero-click currency conversion

10Global CDN exchange-rate API with local caching

20All processing local — no accounts, no tracking scripts

30Works across Chromium, Firefox, and Edge

40Smooth on infinite-scroll and SPA pages

50Full source published and open to contributions

Flux converts and localizes prices on any webpage using live exchange rates — no clicks required. It's free, fully open source, and built to work the same way across every major browser.

Live site → Source → Chrome Web Store →

ALSO ON THE BOARD

FIG. 2

SubSync

A Flask + Supabase platform for managing school timetables and teacher absences — a custom fairness engine auto-assigns substitutes based on workload and availability.

Backend: Flask (Python) Data: Supabase Core: Custom FairnessEngine

Source →

FIG. 3

Spottr — Wifi Locator

A community-driven app for finding public WiFi hotspots nearby, with Supabase handling auth and location data.

Type: Web app Stack: JavaScript, Supabase Model: Community-sourced data

Source →

FIG. 4

No Way Out

A Python puzzle-adventure mini-project built on a custom engine — separate modules for maps, puzzles, player state, and level progression.

Language: Python License: GPL-3.0 Structure: engine · puzzles · states · maps

Source →

More drawings on file — additional repositories, issues, and pull requests.

View full archive on GitHub →

SHEET 04

Title block

DRAWN BY

Satya Siba Nayak

SCALE

NTS

SHEET

04 of 04

DWG NO.

PORTFOLIO-01

GITHUB

@Satya-Siba-Nayak ↗

GITLAB

@Satya-Siba-Nayak ↗

nayak-satya ↗

GOOGLE DEVELOPER

SatyaNayak ↗

STATUS: OPEN TO CONNECT