Amirali Lotfi _
Software Engineer | Full-Stack Developer
$ cat summary.txt
Full-Stack Software Engineer with 4+ years of production experience, with growing hands-on depth in backend development and specializing in frontend engineering. Comfortable working across the stack — from building responsive, performant UIs to designing REST APIs, data models, and backend services with frameworks like FastAPI, Express.js, and NestJS. Familiar with system design principles, authentication flows, and database design.
> Front-end Developer
@ AbrNOC (Internationally Cloudzy) · Dubai, UAE
stack: React, Vue, TypeScript, Material UI, Redux, Zustand, React Query, Nx
- -Maintained and optimized legacy Vue 2 control panels serving 100,000+ active users, supporting thousands of daily interactions across critical workflows.
- -Co-designed the architecture for the platform's full rewrite — defining the React/Nx monorepo structure, shared state model, and incremental migration path off Vue 2.
- -Shipped a shared component library consumed by 4 apps in the monorepo; standardized UI across the platform and let feature teams skip 6+ hours of setup per task.
> Software Engineer
@ Payam Pardaz (Synta Startup Co-Founder) · Isfahan, Iran
stack: FastAPI, MongoDB, LangGraph, React, Material UI, Redux
- -Built founding product features: as a member of the founding team, designed and developed the core features from scratch, taking the product from 0 to MVP in under 3 months.
- -Contributed to system architecture decisions that transitioned the project from a rapid prototype to a production-ready environment capable of handling high-concurrency API requests.
- -Architected an LLM-powered assistant using Gemini and LangGraph, enabling complex, multi-turn user requests with a 90%+ accuracy rate in intent recognition.
> Machine Learning Intern
@ Golrang System · Tehran, Iran
stack: Python, Pandas, NetworkX, PM4Py, FastAPI, MySQL
- -Identified and fixed a long-standing logic flaw in the team's main process mining engine by combining techniques from several recent research papers into a hybrid approach.
- -Improved the accuracy of process discovery on real client data, which the team credited as a contributing factor in a follow-on engagement with a major enterprise client.
> Front-end Developer
@ ArioTech · Tehran, Iran
stack: React, Tailwind, Leaflet
- -Partnered closely with the Backend and Algorithm teams to translate high-dimensional data into interactive GIS-based maps, enabling real-time visualization of complex graph networks.
- -Streamlined the rendering of large-scale datasets on the frontend, improving map load times by 40% and ensuring a smooth user experience for data-heavy network analysis.
> Front-end Developer
@ Nireeka · Toronto, Canada
stack: Next.js, Tailwind, Redux, Redux Saga
- -Participated in Vue to Next.js + TypeScript migration of core pages; SSR/SSG implementation lifted Lighthouse from 60 to 95, cut LCP 35%, and drove a 20% gain in organic traffic.
- -Developed a 3D/interactive bike configurator from scratch, allowing users to personalize parts in real-time.
- -Architected and launched a user forum and customer activity dashboard for 10K+ monthly active users, integrating backend microservices over REST and WebSocket to power a real-time activity feed.
> B.Sc. Computer Engineering
[ 2021 - 2025 ]@ University of Isfahan
Software Engineering & Artificial Intelligence
- EnglishProfessional Working Proficiency
- PersianNative
Unwritten Realms
Interactive story-generation platform where NPCs are autonomous LLM agents with their own memory, knowledge, and goals. Built on LangGraph with a REST API for web integration.
Conversational AI for Mental Health Assessment
AI assistant for semi-structured interviews with older adults, dynamically guiding conversations using predefined psychological frameworks.
Expensive — Bill-Splitting Backend
Server application for Expensive, a bill-splitting app. REST API, auth, and data models built with NestJS and TypeScript, containerized with Docker.
ECG Signal Classification with Deep Learning
Built a deep learning model for time-series classification of ECG signals. Achieved strong performance on binary and multi-class tasks.
Hand Gesture Recognition using YOLO
Fine-tuned a YOLO model for real-time gesture detection. Improved accuracy through dataset curation and preprocessing.
Markov Decision Process Solvers
Implemented Value Iteration and Policy Evaluation/Iteration to solve Markov Decision Processes in stochastic environments.
Reinforcement Learning Agents
Implemented SARSA and Q-learning control agents and evaluated their behavior across stochastic environments.