UPI Fraud Detection System
A fraud-detection ensemble applied to 7.2M+ transactions, combining statistical forensics with explainable predictions. Reported results include 97.3% detection accuracy and sub-100ms API latency at 850+ TPS.
Computer science · AI/ML · Chennai, India
I build data-driven software that makes complex evidence usable, from real-time analytics and retrieval systems to explainable machine learning.
Final-year Integrated M.Tech student in Computer Science Engineering (Business Analytics) at VIT Chennai. Exploring quantum information and computing alongside work in machine learning, backend systems and applied research.

Applied work across machine learning, retrieval systems, security and full-stack engineering.
A fraud-detection ensemble applied to 7.2M+ transactions, combining statistical forensics with explainable predictions. Reported results include 97.3% detection accuracy and sub-100ms API latency at 850+ TPS.
Extracts structured facts from financial and macroeconomic PDFs and ties claims back to quotations and PDF coordinates. A deterministic gate checks period, scope, unit and issuer before comparing documents.
Hybrid dense and sparse retrieval using Legal-BERT embeddings with FAISS and BM25 at a 60/40 ratio. Clause-aware chunking, recursive GraphRAG and confidence-scored self-reflection help produce cited answers and reformulate low-confidence queries.
View code
Scores recovery strategies for failed payments with business guardrails. Reported results include 128 passing automated tests against Razorpay test-mode APIs.
A molecular-data ETL and feature-engineering pipeline with tracked model evaluations, typed FastAPI endpoints and a React interface. Includes Intel OneAPI-based molecular property prediction.
View code
A research blockchain framework combining lattice-based cryptography, validator privacy and consensus experiments, with a Sepolia testnet deployment documented in its repository. Post-quantum cryptography protects against quantum attacks; it is not itself a quantum-computing implementation.
View code
A blockchain-based workflow for farm-to-consumer provenance, intended to make agricultural supply-chain records transparent and tamper-resistant.
View code
A stray-dog care platform connecting field reports, GPS telemetry and operations data. Includes 50+ REST routes and a report-tagging and deduplication workflow.
View code
A quiz platform with code-based joining, live leaderboards and instant scoring, using Supabase realtime and PostgreSQL row-level security.
Open live site
Commercial websites for sport, retail and community organisations.
A multi-sport training academy website with athlete-registration flows and Razorpay payment integration.
Visit website
A gift-store catalog with filtering, search-friendly pages and WhatsApp lead generation.
Visit website
A charity website with structured content and donation-focused information.
Visit website
An e-commerce storefront for sports nutrition and health supplements, with product categories and an online shopping catalog.
Visit website
May–July 2025 · Chennai. Built Python/FastAPI services for energy time-series ingestion, anomaly detection and load forecasting, plus a React/Redux analytics dashboard. Reported results include a 30% reduction in operational analysis time.
June–July 2024 · Chennai. Designed an indexed SQL schema and metadata lineage layer for manufacturing data; developed an edge-optimized YOLOv8 pipeline for classification and evaluation.
Campus leadership, research initiatives and competitive engineering.
Research publication: Enhancing UWSN Security using QKD-enabled Acoustic-Optical Hybrid Communication. I am now looking to develop deeper hands-on work in quantum information and computing. My projects span AI, data, software engineering and post-quantum blockchain security. They are not presented here as quantum-computing implementations.
Interested in research or engineering internships in quantum information, AI/ML or backend systems. Contact me at aashik1701@gmail.com.
GitHub · LinkedIn · Hugging Face