Mohammad Kaamil Mirza
AI Systems Developer & Full Stack Engineer
▸ Summary
Software Engineer with 3 years building production AI and backend systems across Python, TypeScript, Azure, and AWS. Shipped hybrid-retrieval RAG, multi-agent LLM orchestration, voice, payment and reconciliation platforms, and high-throughput ETL for enterprise, healthcare, real-estate and e-commerce clients. Currently completing a Master's in IT Security (AI in Security) at Ontario Tech University.
⚡ Work Experience
AI Systems Developer
@AFM Industries · Whitby, ON
- →Lifted Recall@10 from 71% to 94% with a hybrid-retrieval RAG (vector + BM25, RRF fusion) over 1.2M chunks.
- →Cut p95 query time 2.1s → 340ms by redesigning PostgreSQL schemas across product, vector, and ETL layers.
- →Routed tool calls across Azure AI Search and PostgreSQL nodes via multi-agent orchestration in LangGraph.
- →Cut new-module setup to under a day with a FastAPI scaffold wired to Azure AI Search and GitHub PR checks.
- →Unified reporting on a normalized SQL warehouse via Power BI dataflows and Azure Functions pipelines.
- →Chose Python/FastAPI over Azure Logic Apps, Copilot Studio, and Power Automate by benchmarking cost and vendor lock-in with engineering stakeholders.
Software Engineer (Python, Full-Stack) · Contract
@Independent Contractor · Remote
- →Eliminated dev/prod drift for a client platform team by migrating services to permanent infra via GitLab CI/CD.
- →Unblocked nightly runs by refactoring Airflow DAGs and AWS task groups, fixing Gunicorn auth failures.
- →Blocked breaking changes by automating API contract checks via Swagger and Git hooks across cross-functional teams.
- →Surfaced silent write failures by extending storage observability across AWS EFS, rclone, and HDFS.
- →Upgraded Django 5.2 and Node.js 22 and enforced ruff/Pylint monorepo-wide.
Software Engineer
@Xelpmoc Design & Tech. · Hyderabad, India
- →Reduced pipeline runtime 75% by engineering Python/Spark ETL processing 10M+ records daily.
- →Reconciled 10,000+ financial files daily at 92%+ accuracy by building a Python/AWS reconciliation system with automated exception handling.
- →Selected the production retrieval stack by architecting a configurable RAG platform in LangChain and benchmarking 25+ pipelines on latency and precision.
- →Extracted 5,000+ documents daily at 92% accuracy by building OCR pipelines with PyTorch and LayoutLM; sustained 10K+ messages/sec on Kafka/Redis microservices.
SDE-1, Full-Stack Developer
@Lumara Jewels · Hyderabad, India
- →Automated captures, refunds, settlements, and reconciliation by integrating payment gateway APIs.
- →Secured payment callbacks by designing REST APIs with S2S verification and signed webhooks.
- →Handled concurrent payment failures safely with event-driven services and idempotent retries.
- →Served faceted search over 5,000+ SKUs at sub-20ms by engineering Elasticsearch indices and query tuning.
- →Kept search in sync by deploying MongoDB change-stream pipelines with bulk indexing and retries.
- →Cut UI iteration cycles 50% by architecting server-driven UI (SDUI) with a real-time template engine.
- →Replaced manual discount entry with a promotion engine supporting stacked rules and timed activations.
- →Reduced cold-start latency 60% by optimizing Redis caching for product filters.
SDE Intern, Full-Stack Developer
@Lumara Jewels · Hyderabad, India
- →Priced 5,000+ SKUs at sub-100ms with a TypeScript/Node.js engine over MongoDB.
- →Removed hardcoded pricing by designing modular adapters for material rates, weights, and tax rules.
- →Delivered 30+ invoice and transactional mail templates by wiring AWS S3 storage to Resend.
- →Lifted organic CTR 35% and Lighthouse SEO to 100 with SSR, canonical URLs, and JSON-LD in Next.js.
- →Cut escaped bugs 65% by reaching 90%+ test coverage across 50+ components with Jest, RTL, and Cypress.
🎓 Education
Masters of Information Technology Security — Artificial Intelligence in Security (MITS-AIS)
Ontario Tech University, Oshawa, ON — GPA: 4.0/4.3
Coursework: Machine Learning, Network Data Analysis, Secure Software Systems, Cryptography and Secure Communications, Attack and Defense, Security Policies and Risk Management, Law and Ethics of IT Security
B.Tech. Computer Science and Information Technology
MLR Institute of Technology (Affiliated to JNTU Hyderabad), Hyderabad, India — CGPA: 7.66/10
Coursework: Data Structures, Design and Analysis of Algorithms, Database Management Systems, Operating Systems, Computer Networks, Object-Oriented Programming (Java), Artificial Intelligence, Machine Learning, Data Warehousing and Data Mining, Cloud Computing, Software Engineering, Cryptography and Network Security, Probability and Statistics
🚀 Projects
Razorpay-MCP Server — NPM Package [Open Source]
- →Authored an open-source NPM package implementing an MCP server for AI assistants in TypeScript/Node.js.
- →Exposed payments, orders, and settlements via REST APIs with pagination and structured error handling.
- →Supported flexible deployment by implementing stdio and SSE transports with Docker.
Companion — Note-Sharing Mobile Application
- →Reached 2,000+ downloads and 200 DAUs by launching a campus app on Flutter, Node.js, and MongoDB.
- →Enabled 500+ note shares and real-time chat with Firebase and WebSockets.
💻 Technical Skills
Languages
AI & ML
Backend & Frameworks
Cloud & DevOps
Data & Storage
Payments & Integrations
Testing & Observability
🏆 Achievements
Hack the 6ix 2026 — Winner
Jul 2026, University of Toronto — 1st of 133 teams. Built Tabi, a LangGraph multi-agent travel agent with a supervisor node self-scoring 4 candidate replies per turn; post-trained Qwen3.5-4B via SFT → distillation → GRPO to 100% schema validity. Stack: Gemini, MongoDB Atlas, FastAPI, Docker.
Google Hack4Change — Winner
Aug 2023, T-Hub / Google India / ISB. Built a Python/TensorFlow crop monitoring model at 92% accuracy using satellite imagery.
Major League Hacking — 2nd Place
2023. Built a full-stack solution with React, Node.js, and MongoDB for cognitive retraining of children with disabilities.
📄 Publications
Research Paper
A Mobile Application for Crop Optimization using Satellite Imagery and Disease Detection using AI
IJRASET, ISSN 2321-9653, Vol. 12 Issue IV, Apr 2024
Current Status
Open to opportunities 🚀
Based in Toronto, ON · Remote / Hybrid