Hero

Hello,
This is DISHU BANSAL , I'm a Professional AI Engineer and Software Engineer.

name:'Dishu Bansal',
skills:['Python', 'Java', 'JavaScript/React/TypeScript', 'SQL', 'Cross-platform Development(Desktop/Web/Mobile)', 'Artificial Intellignce/Machine Learning', 'Langchain/Llamaindex', 'LLMs (openAI/Gemini/Anthropic/Grok)', 'Cloud (AWS/GCP)'],
hardWorker:true,
quickLearner:true,
problemSolver:true,
hireable:function() {
return(
this.hardWorker&&
this.problemSolver&&
this.skills.length>=5
);
};
};

Who I am?

Hi, I’m Dishu — a software developer with a passion for building smart, useful, and user-friendly technology. I love turning complex ideas into intuitive digital experiences, whether it’s through clean code, automation, or creating products like Dockie, my AI-powered file explorer. My background spans full-stack development, AI integration, and solving real-world problems with practical tools. I’ve worked as a freelancer, an intern at Geotab, and contributed to several personal projects that reflect my curiosity and love for experimentation. When I’m not coding, I’m usually exploring tech trends, sharing insights on LinkedIn, or learning something completely new to push my limits. I value clean design, smart functionality, and products that just work. I’m currently open to freelance opportunities, collaborations, or full-time roles where I can grow and make an impact. Let’s build something amazing together.

Dishu Bansal
Hero
Experiences
Hero

(Jan 2025 - Present)

AI Engineer

Individual Projects

  • Designed and deployed Jarvis, a real-time multimodal assistant with voice interaction and memory persistence using Gemini 1.5 Flash and Neo4j GDS, improving recall accuracy by 85%.
  • Built Dockie, a smart file explorer using EasyOCR, FAISS, and SQLite, enabling natural language file retrieval with 92% relevance match on semantic search queries.
  • Orchestrated a hybrid memory system with vector search (FAISS) and graph memory (Neo4j) to support dynamic multi-turn conversations and file lookup.
  • Implemented parallelized OCR processing with multithreading and chunked document pipelines, reducing document processing latency from 12s to under 3s.
Hero

(May 2024 - Dec 2024)

Software Engineer Intern

Geotab

  • Maintained and enhanced MyGeotab SDK and REST APIs built with C# .NET MVC, supporting 100+ object models while contributing to frontend features using React and JavaScript.
  • Improved system responsiveness by converting 1000+ lines of synchronous code to asynchronous workflows, resulting in 20% faster execution and 10% lower resource usage.
  • Simulated production-like environments using GCP VMs and logging tools to debug and resolve 10+ high-priority issues, directly impacting user experience for enterprise clients.
  • Implemented rate limiting across 200+ REST API endpoints, reducing server load by 25% and improving API performance by 10%, validated with unit and integration tests using XUnit, Moq, and Selenium.
  • Led incident response for critical on-call issues, independently mitigating downtime within 15 minutes for customer-facing services used by 10K+ users.
Hero

(Jan 2023 - Oct 2024)

Lead Software Engineer

iBuy

  • Led the architecture of a cashback rewards platform using Node.js and MongoDB that validated user-uploaded receipts via OCR and awarded incentives based on spend thresholds.
  • Integrated a robust receipt verification pipeline using EasyOCR and OpenCV, resulting in a 70% reduction in fraudulent submissions.
  • Built an admin dashboard with real-time analytics on user spending patterns using React and Chart.js, improving decision-making speed by 40%.
  • Deployed platform on scalable cloud infrastructure using Render and Docker, supporting 10x user growth without downtime.
Hero

(May 2022 - April 2023)

Software Engineer Intern

SOTI

  • Maintained and extended MobiControl server handling 5M+ devices daily, using C# .NET Core MVC, ASP.NET, C++, and AWS-hosted infrastructure for deployment at scale.
  • Worked in an agile SDLC to deliver production-ready features and iterated on feedback, contributing to internal release documentation and bug fix summaries across sprints.
  • Collaborated with Android engineers to ship features such as VPN control and remote device lock, while independently delivering UI improvements and bug fixes.
  • Improved system stability by debugging SQL logic with MySQL and MS SQL Server, resolving 5+ bugs and fixing 3 flaky tests affecting core functionality.
  • Increased test reliability by 20% by designing and executing over 90 tests, including 50+ unit tests, 30+ integration tests, and 10+ load tests.
Hero

(April 2022 - September 2020)

Lead Software Engineer

AttendIt

  • Developed a secure attendance tracking system with face recognition (FaceNet), device fingerprinting, and GPS fencing to eliminate proxy check-ins.
  • Used OpenCV and Dlib for face verification and TensorFlow Lite for on-device inference, achieving 98.6% face match accuracy.
  • Built RESTful APIs in Flask with JWT-based auth for multi-device tracking and real-time alerts, reducing false attendance by 80%.
  • Integrated MongoDB and AWS S3 for scalable data storage and biometric logs, handling over 100k records with zero data loss incidents.
Skills
PROJECTS

Dockie – AI-Powered Smart File Explorer

name:'Dockie – AI-Powered Smart File Explorer',
tools: ['Python', 'Flutter', 'REST APIs', 'LLM', 'OCR', 'Gemini', 'FAISS', 'SQLite', 'Multithreading', 'File I/O],
myRole:AI Engineer,
Description: Dockie is an AI-driven desktop application that transforms the way users interact with their files. It performs OCR on documents, summarizes content using LLMs (Gemini), and enables natural language search across the user's local file system. Built with speed in mind, Dockie supports multithreaded processing and a vector-based semantic search engine using FAISS. Designed to act like a personal assistant, it makes file retrieval effortless and intelligent.,
};

Jarvis – Real-Time AI Voice Assistant

name:'Jarvis – Real-Time AI Voice Assistant',
tools: ['Langchain', 'Llamaindex', 'Whisper (STT)', 'Piper (TTS)', 'Gemini', 'WebSockets', 'Python', 'Neo4j (Graph Memory)', 'Time-Series Logging],
myRole:AI Engineer,
Description: Jarvis is a fully voice-controlled personal AI assistant built with a focus on personalization and long-term memory. It streams microphone input to Whisper for speech recognition, queries Gemini Flash for intelligent responses, and uses Piper for low-latency TTS. Memory is persistently stored in a Neo4j graph database, enabling complex recall, context tracking, and deep reasoning over time. Jarvis orchestrates real-time input/output while maintaining a scalable, modular architecture — a research-level AI system in the making.,
};

iBuy – Cashback Loyalty Platform

name:'iBuy – Cashback Loyalty Platform',
tools: ['Java', 'Flutter', 'Python', 'Real time', 'GCP', 'Mobile App Development', 'Cross-platform', 'React.js],
myRole:Lead Full Stack Developer,
Description: iBuy is a cashback platform where users commit to spending a target amount at a store. They upload receipts, which are automatically parsed using OCR for verification. The platform tracks progress and issues cashback upon completion. Built with robust back-end logic and an intuitive interface, iBuy blends AI utility with real-world finance workflows — optimizing for fraud prevention and user retention.,
};

AttendIt – Secure AI-Based Attendance System

name:'AttendIt – Secure AI-Based Attendance System',
tools: ['Android (Java/Kotlin)', 'Firebase', 'Face Recognition', 'GPS', 'REST APIs', 'AWS', 'Amplify],
myRole:Lead Full Stack Developer,
Description: AttendIt is a mobile attendance solution combining face recognition, GPS validation, and device-locking to ensure secure, tamper-proof attendance. Designed for educational institutions and corporate environments, the app enforces strict authenticity checks, making false check-ins virtually impossible. It demonstrates applied machine learning and mobile-first engineering in a privacy-conscious framework.,
};

Project Jarvis – AI Traffic Violation Detector

name:'Project Jarvis – AI Traffic Violation Detector',
tools: ['Python', 'OpenCV', 'YOLOv5', 'Computer Vision', 'Flask', 'Android (Java/Kotlin)', 'Firebase],
myRole:Lead Machine Learning Engineer,
Description: This project uses deep learning to analyze dashcam footage and detect traffic rule violations such as red-light jumps or illegal turns. Built using YOLOv5 for object detection and DeepSORT for multi-object tracking, it processes video streams frame-by-frame and flags violations based on movement patterns. It demonstrates real-time video analytics, model integration, and deployment-ready computer vision for smart transport systems.,
};

Signful – Real-Time Sign Language Translator

name:'Signful – Real-Time Sign Language Translator',
tools: ['Python', 'OpenCV', 'MediaPipe', 'TensorFlow/Keras', 'Flask', 'Android (Java/Kotlin)', 'Firebase],
myRole:Lead Machine Learning Engineer,
Description: Signful is a real-time ASL-to-text translator that uses computer vision and deep learning to interpret hand gestures through webcam input. Built using MediaPipe and a custom-trained neural network, it processes frames live and outputs human-readable text to improve accessibility. This project showcases my ability to deploy ML models in production environments with live user interaction and performance optimization.,
};
Hero
Educations
Hero

2020 - 2025

Bachelor Degree

Toronto Metropolitan University (formerly Ryerson University)

  • Graduated with distinction from a competitive co-op program, recognized on the Dean’s List for two academic years due to consistent A+ performance in core computer science courses.
  • Developed a strong foundation in Data Structures, Algorithms, and Object-Oriented Programming, with deep practical exposure in C, Java, and Python.
  • Mastered Linux-based development environments, terminal workflows, system calls, and memory management through courses in Operating Systems and Computer Organization.
  • Built and deployed native mobile applications through Google's Android Development specialization, with hands-on projects using Java and XML layouts.
  • Completed intensive coursework in Computer Networks, Artificial Intelligence, Web Systems, and Security, leading to the ability to design and analyze scalable software systems end-to-end.

Contact with me

If you have any questions or concerns, please don't hesitate to contact me. I am open to any work opportunities that align with my skills and interests.

dishu.bansal@torontomu.ca

+16476161501

Toronto, Ontario, Canada M5B 1Y6