Ishan Rahul Surdi — Computer Science

About

A short introduction

I'm a Computer Science undergraduate at MIT World Peace University, Pune, graduating in 2027. My current work spans quantum-inspired optimization, AI security, and open-source scientific/engineering software.

My research interests sit at the intersection of machine learning and careful measurement: benchmarking algorithms against honest baselines, studying failure modes in applied AI systems, and building tools that make technical claims easier to reproduce.

AI / ML Quantum Computing Cloud Computing Cybersecurity Research Full-stack Development Data Science

Education

Academic background

MIT World Peace University

Computer Science

Current CGPA: 9.03. Coursework and project work focused on computer science foundations, software engineering, data structures, machine learning, and applied systems.

2024 — 2027
Pune, India

Experience

Where I've worked

Research internships and engineering roles, most recent first.

Summer Research Fellow

FOSSEE, IIT Bombay

2026 — Present
Remote

CurrentResearchOpen SourceRemote

Contributing to 3psLCCA, an open-source life-cycle cost analysis tool for three-span PSC bridges.
Debugged and rebuilt the PyLaTeX report-generation pipeline: fonts, tables, escapes, and annotated PDF charts.
Reviewed and fixed contributor pull requests on the Osdag Bridge codebase, including import and packaging issues.
Designed the LaTeX/TikZ report cover system used in generated engineering reports.

AI Security Research Intern

RV University · Prof. Manish Kumar

2026 — Present
Remote

CurrentResearchAI Security

Researching intelligent detection of QR-code phishing (quishing) attacks.
Working across URL feature analysis, visual QR inspection, and ML-based classification.

Quantum Computing Research Intern

DES Pune University · Dr. Prathamesh Lahande

2025 — Present
Pune, India

CurrentResearchQuantum

Built a rigorous benchmark of quantum-inspired metaheuristics (QIEA/MQGA, MQPSO, RQWOA) against classical analogs.
Designed ablation studies isolating the contribution of each "quantum" operator, with statistical significance testing.
Found that several quantum-inspired operators are statistically inert or harmful versus tuned classical baselines — and framed the thesis around that honest result.

Software Engineering Intern

Infosys Springboard

2025
Remote

CompletedEngineeringRemote

Worked on distributed microservices and a real-time computer-vision inference pipeline.
Gained hands-on experience with service communication, deployment, and inference latency trade-offs.

Research

Publications & work in progress

Papers under review and published work.

A Hybrid XGBoost-CNN-1D Framework for Workload-Aware HPC in Cloud System

2026
Research paper

Under reviewHPCCloud schedulingML systems

Under the guidance of: Dr. Prathamesh Lahande, DES Pune University
Submitted to: International Journal of Computing Journal

Combines XGBoost and a 1-D CNN with recall-union fusion to flag high-work jobs from workload traces and route them across cloud VMs.

TAC-Net: Transition-Aware Adaptive Computation for Efficient Tabular Neural Networks

2026
Research paper

Under reviewDeep learningEfficient inferenceTabular ML

Under the guidance of: Dr. Sachin Bhoite, Dr. Vishwanath Karad MIT World Peace University
Submitted to: Inteligencia Artificial Journal

A learned gate decides, layer by layer, whether deeper computation would fix or harm the current prediction. Preserved accuracy in 13 of 15 runs while cutting executed computation by 36–55%.

Prediction-Guided Rebuilding of Dynamic Vector Indexes for AI Agent Memory

2026
Conference paper

SubmittedAI agentsVector searchAgent memory

Under the guidance of: Dr. Sachin Bhoite, Dr. Vishwanath Karad MIT World Peace University
Submitted to: 2027 International Conference on Emerging Smart Computing and Informatics (ESCI) Conference

Are quantum computers a threat to privacy?

2023
Journal article

PublishedQuantum computingPrivacyCryptography

Published in: International Journal for Research in Applied Science and Engineering Technology (IJRASET) Journal

DOI 10.22214/ijraset.2023.55568

Projects

Selected work

A few projects I've built — research prototypes, hackathon systems, and tools.

EpiSentinel

Disease outbreak forecaster combining Neural ODEs with graph attention networks over data from 201 countries.

Finalist at the IIT BHU hackathon.

PythonPyTorchNeural ODEGNN

GitHub Live demo

AegisKYC

Multilingual OCR pipeline for detecting fraudulent identity documents in KYC workflows.

98.5% accuracy on the internal evaluation set.

PythonOCRScikit-learn

GitHub

SkyQuery AI

Retrieval-augmented generation pipeline grounded in a Neo4j knowledge graph, with structured query decomposition.

Graph-grounded answers instead of free-floating LLM output.

PythonNeo4jRAGLLMs

GitHub

SDK-Sovereign

Multi-agent OpenEnv environment where a Lead and an Auditor agent coordinate over seven turns to migrate code off a cut-off vendor SDK onto a sovereign allow-list.

Live on Hugging Face Spaces with a playable demo and a reproducible training notebook.

PythonMulti-agent RLOpenEnvHugging Face

GitHub Live demo

E-DERA

Financial risk scoring platform with explainable features and a clean analyst-facing dashboard.

End-to-end: ingestion, scoring, and reporting.

PythonPandasStreamlit

GitHub

Image Quality Analyzer

Full-stack app that scores uploaded images 0–100 and flags blur, noise, compression, and under/over-exposure using computer vision and a trained classifier.

Trained on BSDS500 and KADID-10k; deployed with a public API.

PythonFastAPIOpenCVScikit-learn

GitHub Live demo

Activity