👋 Hey there, I'm

Aditya Singh

|

3rd-year CS student at BITS Pilani specializing in Computer Vision and Deep Learning. Hackathon finalist with 99%+ model accuracy. Research work in stereo vision and ML. Former intern at Beckn Protocol (Digital Energy Grid) and bug bounty contributor who saved ₹4L+ for FamPay.

5+ Hackathon Finalist
10+ ML Projects
99% Best Accuracy
View Projects Get in Touch
Aditya Singh

01. About Me

I'm a 3rd-year Computer Science student at BITS Pilani (BITSAT Top 1%), pushing the boundaries of Machine Learning and Computer Vision. My work spans from hackathon finalist projects to research-grade computer vision systems.

Currently working on stereo vision for agricultural automation under Prof. Gopal Singh Phartiyal. Finalist at the Eightfold AI × ArIES Hackathon (IIT Delhi) with CodeSage, the world's first voice-powered AI technical interviewer.

Former intern at Beckn Protocol's Digital Energy Grid where I generated 10M+ test data points for search engine validation. As a bug bounty researcher, I discovered a critical infinite loop vulnerability in FamPay that saved the company ₹4L+ in fraudulent card orders.

Finalist - Eightfold AI Hackathon (IIT Delhi)
Research under BITS Pilani Faculty
Saved ₹4L+ via Bug Bounty (FamPay)
Top 5% Kaggle

02. Experience

May – Aug 2025

Software Developer Intern

Beckn Protocol - Digital Energy Grid, Bangalore

Contributed to the Digital Energy Grid (DEG) project - a unified, interoperable energy infrastructure. Generated 10 million+ synthetic energy profiles using Python Faker for search engine testing. Created geo-coordinated data for city-specific testing and worked with Strapi, Postman, and Docker for Beckn system integration.

10M+ Test Data Points
Search Engine Validation
Python Faker Strapi Docker Postman Beckn Protocol
View Project →
Oct 2024 – Jan 2025

Bug Bounty Researcher

Independent (FamPay)

Discovered critical infinite loop vulnerability in FamPay's card ordering system that allowed users to order unlimited free Rupay debit cards despite single payment. Identified flawed for-loop logic enabling multiple cards to connect to single UPI IDs. Estimated financial impact: Saved ₹4 lakh+ in fraudulent orders and compliance costs.

₹4L+ Saved
Critical Severity
Security Research Payment Systems Logic Flaws UPI Security

03. Technical Arsenal

ML & Deep Learning

PyTorch TensorFlow Scikit-learn Transformers XGBoost Keras

Computer Vision

OpenCV Stereo Vision GANs EfficientNet DINOv2 FAISS

Audio Processing

WavLM ASR Signal Processing Deepfake Detection LFCC

Development

Python JavaScript React Flask FastAPI Git

Tools & Platforms

Linux Docker Strapi Postman Google Colab Jupyter

Security & Other

Bug Bounty Security Research Web Scraping Data Generation

04. Featured Projects

Hackathon Finalist

CodeSage: AI Voice Technical Interviewer

World's first voice-powered AI technical interviewer that addresses the $12B technical interview market. Built with Google Gemini 2.0 Flash, featuring real-time voice interaction with 2-second response time, 100-point evaluation system, and enterprise-grade security suite with multi-modal cheating detection.

Finalist - Eightfold AI (IIT Delhi)
2s AI Response Time
Saves $2,847/Interview
Gemini 2.0 React 18 Flask TensorFlow.js Monaco Editor
Code Enhanced Version
CodeSage Project
Research Project

Stereo Vision for Agricultural Automation

Research-grade depth estimation and fruit ripeness analyzer using stereo vision and feature learning. Developed under Prof. Gopal Singh Phartiyal at BITS Pilani. Novel fusion of LCDM + MBM models achieving 94-99% depth accuracy with real-time embedded inference for smart-farming ripeness detection.

BITS Pilani Research
Real-time Processing
94-99% Depth Accuracy
OpenCV Stereo Vision SVMs NumPy Python
Code
Depth Estimation Project
Hackathon Finalist

CloudFlux AI: Multi-Cloud Orchestration

Intelligent data management platform for NetApp Hackathon that automatically classifies data into HOT/WARM/COLD tiers using ML. Features stunning React UI with particle effects, real-time analytics, and multi-cloud support (AWS, Azure, GCP). Achieves 40-60% cost reduction with sub-100ms classification.

$531K Annual Savings
<100ms Classification
Multi-Cloud Support
React FastAPI Material-UI ML Classification
Code Full Version
CloudFlux Project

More Projects

Audio Deepfake Detector

4th place in Comsys Hackathon-6. Dual-branch CNN fusing LFCC spectrograms with raw-waveform speaker embeddings. Achieved Macro-F1 = 0.9991 with Tesla M10 optimization and ONNX export for deployment.

4th Place F1: 0.9991
CNN LFCC ONNX

Gender Classification & Identity Recognition

Dual-task deep learning solution achieving 90%+ balanced accuracy for gender classification and 98-99% accuracy for identity recognition under distortions.

90%+ Accuracy DINOv2 + FAISS
EfficientNet DINOv2 FAISS

Smart Product Pricing Challenge

ML competition solution with 32 engineered features and ensemble models. Achieved 66 SMAPE on 75K samples with 3-minute runtime.

66 SMAPE 75K Samples
RandomForest XGBoost Feature Engineering

Advanced Web Scraper with AI

Intelligent web scraping tool with multiple AI providers including free Ollama. Exports to Word, CSV, JSON with sentiment analysis.

Multi-AI Selenium
Python Ollama BeautifulSoup

Crypto Sentiment Trader

Smart trading pipeline combining Reddit sentiment, news analysis, and ML. Achieved 78% accuracy with 1.42 Sharpe ratio using BERT/VADER fusion.

78% Accuracy 1.42 Sharpe
XGBoost VADER Backtrader

05. Get In Touch

Let's Build Something Amazing

I'm currently seeking research internships at top institutions (EPFL, Max Planck) and ML engineering opportunities. Whether you want to collaborate on a project, discuss research ideas, or just say hi, my inbox is always open!

Education BITS Pilani (3rd Year CS)
Location India
Email spirokinetics@gmail.com
Phone +91-9257473483

GitHub

@Adityas221b

LinkedIn

Connect with me

Kaggle

Top 5% Competitor

Email

Drop a message