Ojaswi Singh — AI Researcher & CS Undergraduate

About

"Pre-final-year B.Tech CSE (AI/ML) student with interests in Machine Learning, Deep Learning, Computer Vision, and Generative AI."

Currently working as an AI Research Intern, developing a research project on single-image depth-from-defocus estimation using blur-aware transformer architectures. Also built an AI-based fake news and propaganda detection system using NLP and machine learning, achieving over 95% classification accuracy. Beyond research, serves as Operations Manager at Microsoft Tech Community, contributing to technical events, community initiatives, and student engagement — while continuing to strengthen DSA, AI/ML, and software development skills.

Education

Amity University

2024–2028

B.Tech, Computer Science · Minor in AI & Machine Learning

Noida, Uttar Pradesh

CGPA: 9.44/10 (current)

Birkbeck, University of London

Apr–May 2026

Study Abroad Programme

London, United Kingdom

Coursework in Business Management and New Leadership in Cultural Contexts (NLCC)
Collaborated on a comparative Country Report; Best Research Paper Award at the associated conference

Experience

AI Research Intern · Amity University

JAN 2026 — PRESENT · NOIDA, UTTAR PRADESH

Selected among 30 students from 600+ applicants for a university-led AI research internship.
Developing a transformer-based framework for single-image depth estimation from defocused images.
Implemented preprocessing, training, and inference pipelines using Python, TensorFlow, and OpenCV.
Conducted model evaluation and performance analysis for monocular depth prediction tasks.

Operations Manager · Microsoft Tech Community

SEP 2025 — PRESENT · NOIDA, UTTAR PRADESH

Promoted from Core Member to Operations Manager for contributions to community operations and event management.
Led operational planning and coordination for technical workshops, hackathons, and student engagement initiatives.
Worked with organizing teams to streamline event execution, logistics, and participant communication.
Supported outreach programs focused on AI, cloud computing, and emerging technologies.

Research Spotlight

Flagship Project

Blur-Aware Transformer-Based Monocular Depth Estimation

A comparative study of seven CNN and Transformer backbones for estimating a dense depth map from a single RGB image, trained on NYU Depth V2 under identical conditions with a shared decoder — isolating the encoder as the only variable.

Swin Transformer MIT License ★ 1

Swin-Tiny

0.503

ViT-Small

0.439

ResNet18

0.428

ResNet50

0.422

VGG16

0.395

MobileNetV3

0.355

ConvNeXt-T

0.333

δ1 accuracy (higher is better) · pilot run, 240 train / 60 val samples, 224×224 input, frozen ImageNet-pretrained encoders, shared trainable decoder.

Swin-TinyLowest MAE, RMSE & AbsRel

ResNet18Fastest — 322.8 FPS

MobileNetV3Lightest deployable option

Base research paper: Eigen, Puhrsch & Fergus (2014), Depth Map Prediction from a Single Image using a Multi-Scale Deep Network.

View repository & full results →

Projects

Natural Language Processing

Fake News & Propaganda Detection

NLP classifier separating credible reporting from propaganda-risk content.

TF‑IDF, Logistic Regression & Linear SVM on 72,000+ articles · Streamlit app for real-time credibility scoring.

Scikit-LearnStreamlitPandas

View repository →

Biomechanics · Signal Processing

Gait Cycle Segmentation

vGRF-based gait cycle segmentation pipeline, IIT Kharagpur dataset.

Ground-reaction-force gait cycle detection across multiple subjects (Sub01–Sub05, healthy & affected groups).

PythonCythonNumPy

View repository →

CodeAlpha AI Internship

Three-Project AI Toolkit

A trio of applied AI tools built during the CodeAlpha internship, each shipped with a Gradio interface.

Object Detection & TrackingYOLO11n + ByteTrack

University FAQ ChatbotTF‑IDF + Cosine Sim.

Language Translation ToolDeep Translator

Multi-object tracking with persistent IDs, a 35-FAQ university support bot, and an auto-detect translation interface — all runnable from Google Colab.

OpenCVGradioColab

Browse all repositories →

Skills & Certifications

Advanced — used daily in research and projects Intermediate — working knowledge, still deepening

Programming Languages

Python
C++
Java
C

Core Computer Science

Problem Solving
Data Structures & Algorithms
Object-Oriented Programming

AI & Machine Learning

Computer Vision
Deep Learning
Machine Learning
Natural Language Processing
Generative AI
Prompt Engineering

Libraries & Frameworks

TensorFlow
OpenCV
Scikit-Learn
Pandas
NumPy
Gradio
Streamlit

Tools

Git

GitHub

VS Code

Google Colab

Jupyter

Languages

English — ProfessionalHindi — NativeSpanish — Conversational

Soft Skills & Leadership

Leadership
Team Management
Strategic Planning
Operations Management
Operational Execution
Event Management
Public Speaking & Anchoring
Professional Communication
Creativity & Innovation
Public Relations & Outreach
Graphic Design

Certifications & Badges

AI Skills Fest 2026

Microsoft · via Credly

Forward Program

McKinsey.org · via Credly

NVIDIA DLI — Fundamentals of Deep Learning Google Generative AI & Prompt Engineering Specialization (4-course)

Honors & Awards

1st Position · 6 Aug 2026

Amity Ideathon 2026 — AI, Digital Technologies & Smart Systems

With teammates Mahatva Goel & Arihant Jain · powered by IEEE Region 10 · awarded on World Entrepreneurship Day

Ojaswi Singh's team receiving the trophy on stage at Amity Ideathon 2026 — On stage, 1st position

Close-up of the Amity Ideathon 2026 1st position trophy — The trophy

Official Certificate of Merit for Ojaswi Singh, Amity Ideathon 2026 — Certificate of Merit

Best Research Paper · 11 May 2026

"Smart City Technologies in London: Role of AI and Digital Innovation"

International Conference on Global Synergy: Shaping New-Age Management and Enterprise Development with Disruptive Technology · Amity University, London · alongside the Birkbeck study abroad programme