Tasnim
Ashraf.
AI Systems Engineer building agentic infrastructure & production-grade backend systems.
I engineer backend systems and AI agents that turn unstructured, ambiguous input into structured, verifiable output — from LLM-driven parsing pipelines to production-grade APIs. Currently building the Java/Spring Boot backend parser for SysML v2 inside an agentic IDE at SysModeler.ai.
I bridge AI research and product engineering — turning models into systems people actually use.
My work spans machine learning models and responsive full-stack applications. With 8+ hackathon victories, I rapidly prototype and ship — from AI-powered platforms to real-time collaborative tools and payment-integrated products.
This blend of research depth and shipping speed lets me deliver intelligent automation and scalable software that drives measurable results.
AI Systems & Backend Engineer
- —Building the core Java/Spring Boot backend parser that converts SysML v2 textual notation into structured, validated system models
- —Designing agentic pipelines that let an AI-native IDE generate, refine, and reason over SysML v2 models from natural-language and document input
- —Architecting scalable, low-latency backend services and APIs powering real-time collaboration for Model-Based Systems Engineering (MBSE) workflows
Featured projects
Inherit — AI-Powered Coding Education
Educational platform pairing curated video lessons with AI-powered learning roadmaps, time-based coding quests, a real-time collaborative editor, and intelligent live code review.
DevDiscuss — AI Developer Community
Community platform with AI-assisted problem-solving, smart search, code-snippet sharing, and user analytics with exportable PDF reports.
অক্ষর (Okkhor) — Bengali RAG Platform
RAG-based learning platform on LangChain + Pinecone delivering contextual Bengali content with OpenAI integration — a 95% accuracy improvement.
FaceNet Real-Time Attendance
ML attendance system using FaceNet (Inception-ResNet-V1) with MTCNN detection and VGGFace2 weights for real-time student identification.
CrackEd — Admission Prep Platform
Full educational platform for IUT admission tests: mock & model tests, question banks, tutor management, doubt resolution, and mistake-based analytics.
Ibn Sina — Clinic Management
Healthcare management app with doctor dashboards, patient registration, a prescription system, admin controls, and multi-device network support.
8+ competition wins.
Global 4th & National Champion
HackNSU Season 5
BitFest 2025 Hackathon
4th Data Science Summit
Technocrats V.2
Programming Hero National Hackathon
BitFest 2025 Project Showcase
BitFest 2025 Datathon
AI / ML & Deep Learning
AI Frameworks & Tools
Languages
Data Science
Full-Stack Development
Infra & Workflow
Have a project
in mind?
Available for freelance, consulting, and long-term partnerships — AI solutions, full-stack builds, or hybrid systems that combine both. Let's turn your idea into a production-ready product.