Aditya Sahai

I’m an M.Tech student in Computer Science and Engineering with a strong focus on data integrity, automation, and applied machine learning. I like turning manual, error-prone workflows into systems that are fast, testable, and dependable.

As an Analyst Intern – Financial Engineering (Product Control Quants) at HSBC, I built a high-performance Jenkins automation pipeline that cut data processing and deployment from 4–5 days of manual effort to under 2 minutes. I also architected CI/CD workflows for quantitative tools, replacing legacy manual reporting with automated Python-based validation frameworks to ensure 100% data integrity.

My projects blend deep learning and robust tooling: in WhatDish I leveraged a YOLO-based detection pipeline (mAP50: 95.33%) and delivered production-ready TFLite models, while curating a 100k+ image dataset. I’ve also built a deception classifier (83% accuracy) using text plus strategic metadata and explored specialized simulations like a WiFi 4/5/6 OOP simulator to evaluate throughput and latency.

I back my work with measurable outcomes—publishing in IEEE Xplore (90.77% classification accuracy) and having an accepted CVIP 2025 paper using SOTA YOLO models for fine-grained dish recognition. I’m motivated by building systems that improve velocity without sacrificing correctness.