+--+
|  |
+--+
Profile

Sonit Bahl

Final Year Student

📍 Based in New Delhi, India
💻 Ask me about Machine Learning & Cyber Security
📄 If you're still with me, this resume might just seal the deal.

+--+
|  |
+--+
$git stats --user SonitBahl_
Repos
...
Stars
...
Followers
...
Commits
...
terminal
+--+
|  |
+--+
$ls -la tech/_
C iconC
C++ iconC++
C# iconC#
Java iconJava
Python iconPython
Ruby iconRuby
HTML iconHTML
CSS iconCSS
JavaScript iconJavaScript
React.js iconReact.js
Node.js iconNode.js
TensorFlow iconTensorFlow
PyTorch iconPyTorch
Keras iconKeras
Scikit-learn iconScikit-learn
Langchain iconLangchain
Unity iconUnity
GitHub iconGitHub
Jupyter Notebook iconJupyter Notebook
Docker iconDocker
Streamlit iconStreamlit
Wireshark iconWireshark
Paramiko iconParamiko
LLMs iconLLMs
Penetration Testing iconPenetration Testing
System Hardening iconSystem Hardening
SSH Automation iconSSH Automation
Network Monitoring iconNetwork Monitoring
+--+
|  |
+--+
$mpv --playlist favorites.m3u_
Raabta
0:00 / 0:00
Arijit Singh
▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
Playlist5 tracks
Raabta
...
Tu Hain Toh
...
Chura Liya Hai Tumne
...
Agar Tum Saath Ho
...
Aise Na Mujhe Tum Dekho
...
+--+
|  |
+--+
$ls -l experiences/_
+--
--+
Ericsson

Ericsson

July 2025 - Present

Cybersecurity Intern · Noida, India


• Worked on Unix Systems focusing on OS Hardening, PKI, Encryption Algorithms, Hashing, and TLS/IP configuration.
• Used Nmap, Wireshark, Kali Linux, and EJBCA for network scanning, traffic analysis, and certificate management.
• Performed CIA testing on 5+ internal software systems, identifying and documenting 12+ vulnerabilities.
• Implemented patch updates with the security team, ensuring timely resolution of identified system vulnerabilities.
Click to Learn More
+--
--+
Samsung R&D Institute

Samsung R&D Institute

Sept 2024 - May 2025

GenAI Intern · Bangalore, India


• Reviewed 30+ research papers on ViT, Swin, BEiT, DeiT to identify bottlenecks in fixed-size tokenization approach.
• Designed a two-stage adaptive pruning mechanism that scores and drops low-value tokens, improving accuracy by 15% and reducing training time by 25%.
• Reduced token count by up to 8x via content-adaptive patching, preserving accuracy and improving efficiency.
• Conducted ablation studies on token selection, measuring accuracy, speed, and efficiency on Oxford Flowers-102.
Click to Learn More
+--
--+
Ericsson

Ericsson

May 2024 - Jun 2024

ML Intern · Noida, India


• Built Drishti, a 360° vision headband using Python, OpenCV, and camera modules for real-time object detection.
• Developed a video dehazing pipeline with the Dark Channel Prior algorithm, enhancing visibility in low-contrast footage across 10K+ frames.
• Improved object detection accuracy by 18% by integrating multi-camera inputs on dehazed frames.
• Implemented a LiDAR-based risk detection module with real-time audio alerts using pyttsx3 integration.
Click to Learn More
+--+
|  |
+--+
$find ./projects -type f -name '*.featured'_
Interactive SSH Honeypot
Active

Interactive SSH Honeypot

PythonParamiko+4
Aquatwin
Active

Aquatwin

UnityC#+5
+-----------------------+
|     CONNECT WITH ME    |
+-----------------------+
$ssh-connect --social_