ML ENGINEER · DEV · RESEARCHER

Hi, I'm
Anushka!

›

I'm a 20-year-old IT undergrad and ML Engineer with research interests in AI/ML, focusing on building intelligent systems and full-stack AI applications that solve real-world problems through user-centric solutions. With 3+ years of web development experience, I'm an Amazon ML Summer School 2026 Scholar, have won 3 hackathons through team collaboration, and have published 2 research papers in applied AI/ML.

I am an active member of Google's Women Techmakers and IEEE Robotics & Automation Society and part of the NVIDIA Developer Community, exploring advancements in AIML, and modern computing. My interests naturally gravitate toward anything that adds to the growing body of Human Knowledge. For more details about my experience, check out my resume.

SCROLL

Experience

Professional Journey

Engineering roles, open-source contributions, and community work

AMAZON · REMOTE

Amazon ML Summer School (MLSS) Scholar

JUL 2026 – PRESENT

Selected as one of the scholars for Amazon ML Summer School 2026, a cohort-based program delivered directly by Amazon research scientists and applied ML researchers.

Learning advanced machine learning topics including Supervised Learning, Deep Neural Networks, Dimensionality Reduction, Unsupervised Learning, Sequential Learning, Reinforcement Learning, Generative AI & LLMs, and Causal Inference through live sessions and hands-on modules.

FLYRANK AI · REMOTE

ML Intern

JUL 2026 – PRESENT

Developed end-to-end machine learning pipelines for building and deploying robust ranking models, successfully processing over 100,000+ data points and reducing pipeline execution time by 20%.

Implemented advanced feature engineering techniques, including TF-IDF and word representations, to extract highly predictive signals from 50+ raw dataset attributes, boosting the signal-to-noise ratio by 25%.

Architected predictive models utilizing XGBoost, Random Forest, and PyTorch, achieving a 15% improvement in ranking accuracy (NDCG/MAP) across standard evaluation metrics.

OASIS INFOBYTE (AICTE OIB-SIP) · REMOTE

Data Analytics Intern

JUL 2026 – 1 Month

Engineered robust data cleaning and preprocessing pipelines for real-world datasets containing 50,000+ records (e.g., Retail Sales, Airbnb), handling missing values and outliers utilizing Python and Pandas.

Architected Exploratory Data Analysis (EDA) workflows to extract actionable business insights, generating 7+ interactive visualizations utilizing Seaborn and Matplotlib to map customer demographics and seasonal sales trends.

Developed predictive machine learning models, including Random Forest and Support Vector Machine (SVM) classifiers, achieving over 85% classification accuracy through systematic feature engineering and hyperparameter tuning.

Implemented 3 end-to-end analytical solutions—ranging from linear regression for price prediction to sentiment analysis—optimizing model cross-validation scores by 15% using Scikit-learn.

ZETHETA ALGORITHMS PRIVATE LIMITED · REMOTE

Blockchain Risk Developer

JAN 2026 – MAR 2026

Engineered core decentralized infrastructure for the WorkBridge platform, integrating Web3 wallets to facilitate a secure, low-latency Crypto-Fiat Bridge.

Architected a Stablecoin Stress Tester that continuously monitors on-chain telemetry, tracking collateralization ratios across 4+ major liquidity pools to improve systemic risk assessment.

Developed an automated Solidity vulnerability scanner utilizing static analysis to analyze 10,000+ lines of smart contract code, capable of detecting common Reentrancy and Access Control vulnerabilities prior to deployment.

Implemented an Avellaneda-Stoikov market-making engine to programmatically manage inventory risk, optimizing bid-ask spreads and enhancing high-frequency liquidity provision.

GSSoC '25 · REMOTE

Open-Source Contributor

JULY 2025 – SEPT 2025

Contributed to 4 major open-source repositories, successfully merging 15+ Pull Requests (PRs) focused on core feature additions, dependency upgrades, and critical bug resolutions.

Improved algorithmic efficiency through code refactoring, memory optimization, and complexity reduction across multiple modules.

Collaborated asynchronously with a global network of maintainers, conducting strict peer code reviews, resolving merge conflicts, and ensuring adherence to CI/CD testing pipelines and semantic versioning.

Projects

Deployed Engineering Solutions

Select a project below to explore the architecture and technologies used.

⬇ CLICK ON THE CARDS TO LEARN MORE ABOUT THESE PROJECTS

DeepfakeDetector

React · FastAPI · PyTorch

DCO6 Synthesizer

C++ · JUCE · DSP

iNaturalist Clumper

Python · Spatial Data

Deweathering Engine

React · OpenCV · RPCA

SEE OTHER PROJECTS →

These are just some highlights. Explore all repositories, experiments, and works-in-progress.

VIEW ALL ON GITHUB

Technical Arsenal

Core Competencies

Languages, frameworks, and tools across the full stack and ML pipeline

anushka@dev ~ stack.json

// Technical Stack — Anushka

"role": "ML Engineer / Full-Stack",

"languages": [ "Python", "Java", "C++", "SQL", "JS", "HTML", "CSS" ],

"ml_frameworks": [ "PyTorch", "Scikit-learn", "XGBoost", "LightGBM", "OpenAI" ],

"web": [ "React", "FastAPI", "Flask" ],

"cloud_devops": [ "AWS", "Docker", "Git" ],

"actively_shipping": true

Languages

Python

Java

C++

SQL

JavaScript

HTML5

CSS3

Libraries & Frameworks

Scikit-learn

PyTorch

Pandas

NumPy

XGBoost

Flask

FastAPI

React

OpenAI

Tools & Platforms

Amazon AWS

Streamlit

Git

Jupyter

Power BI

Docker

Research & Publications

Advancing Intelligent Systems

Peer-reviewed research in ML efficiency, deepfake forensics, and sovereign AI

QLoRABITNET 1.58PEFTSovereign AI

Architectural Paradigms for Sovereign AI: Mitigating API Fragility through Quantized Low-Rank Adaptation and 1.58-bit Ternary Models [Officially lived on SSRN]

A rigorous methodology for deploying Sovereign AI through Small Language Models (SLMs) to mitigate API fragility. Investigates the synergy between 4-bit NF4 quantization, native 1.58-bit ternary architectures, and LoRA for specialized domains. Officially lived on SSRN.

93% reduction in inference OpEx450M token TCO break-even

2D FFTResNet-18PyTorchDeepfake

Isolating Synthetic Fingerprints: A Frequency-Domain Approach to Robust Deepfake Detection using 2D FFT and Lightweight CNNs

A novel, lightweight deepfake detection engine isolating synthetic fingerprints in the frequency domain. Employs micro-blur preprocessing to suppress superficial high-frequency noise, followed by 2D FFT to extract spectral magnitude maps.

99.12% on FaceForensics++45 FPS real-time inference

Achievements & Certifications

Milestones & Recognition

Awards, accomplishments, and continuous professional development

Amazon ML Summer School 2026

Selected for the highly competitive Amazon Machine Learning Summer School, gaining in-depth exposure to industry-applied ML concepts, deep learning, and scalable AI solutions directly from Amazon scientists.

IndiaFirst Life (Bank of Baroda)

Unexpectedly shortlisted via Foundit and successfully cleared an interview for a Business Administration/Management role as a 6th-semester undergrad. Secured a 2-month training program with a stipend, leading to a full-time offer but respectfully declined to focus on 6th-semester studies.

SIH '25 Semi-Finalist

'Digi-Pramaan' - Top finalist out of 250+ teams in the college internal round of Smart India Hackathon 2025.

Hackathons Winner

I have won 3 hackathons, collaborating with teams to develop innovative solutions.

HackerRank

Gold Level (5 Star) in Python, Silver Level (3 Star) in Java.

Extracurricular Leadership

Active member of 'Samaritans' social club since first year of my undergrad degree - community initiatives, collaborative events and also in debates and quiz competitions.