Hi, I'm Syed Muhammad Khizer Haider
Software Engineer | AI/ML | Computer Vision | Full-Stack | Mobile (Flutter) | Edge AI | IoT
CS undergraduate at NUST (Class of 2027) and Gold Medalist High Achiever with production-grade experience across AI/ML, computer vision, full-stack web, and cross-platform mobile development. Sole engineer on a NUST-funded confidential drone-based aerial detection system — delivered end-to-end from dataset curation to a fully offline edge inference pipeline on NVIDIA Jetson. NASA Space Apps Challenge Winner 2025.
About Me
Versatile CS undergraduate at NUST and Gold Medalist High Achiever with production-grade experience across AI/ML, computer vision, full-stack web, and cross-platform mobile development. Proven track record of shipping complex systems independently under research and commercial constraints.
Education
National University of Sciences & Technology (NUST) — Bachelor of Computer Science
2023 – 2027
Gold Medalist — High Achiever Award 2025
Experience
Sole engineer on a NUST-funded confidential drone-based aerial detection system. Research assistant on funded computer vision projects, full-stack intern at UpSkillers, and network engineering intern on large-scale government surveillance infrastructure.
Areas of Expertise
AI / ML / Computer Vision
Full-Stack Web Development
Mobile & Edge AI (Flutter)
Databases & DevOps
Embedded & Networking / IoT
How I Build AI Systems
A layered approach to engineering intelligent applications — from data to deployment.
Layer 1
Data Collection & Curation
Collect domain-specific data, perform annotation (including OBB for aerial imagery), resolve class imbalance, and build preprocessing pipelines with geometric and photometric augmentations.
Layer 2
Model Training & Optimization
Train models (YOLOv8, Random Forest, LSTM), conduct ablation studies across resolutions and configs, evaluate with mAP/P/R metrics, and iterate on architecture decisions.
Layer 3
Edge Deployment & Inference
Export to TensorRT/TFLite/ONNX, deploy on edge hardware (NVIDIA Jetson, mobile), benchmark delegates across FP32/FP16/INT8, and build offline inference pipelines.
Layer 4
Application & Integration
Ship production apps — Flutter mobile with native Kotlin plugins, real-time WebSocket streams, ground station systems, and full-stack web platforms with REST APIs.
Technical Skills
Technologies and tools I use to build intelligent, scalable systems.
AI / ML / Computer Vision
Full-Stack Web
Mobile & Edge
Databases & DevOps
Embedded & Networking
Work Experience
Research, engineering internships, and hands-on industry experience.
Research Assistant
NUST Funded Project (Computer Vision — Confidential)
- Sole engineer on a confidential aerial object detection system — full ownership from raw data collection to a production-deployed, fully offline edge pipeline on a DJI Matrice 4 drone platform
- Collected drone-captured imagery; performed OBB annotation for 7 domain-specific classes; resolved class imbalance via targeted re-collection and weighted sampling
- Built a full data preprocessing pipeline: geometric augmentations (flips, rotation, perspective warp), photometric augmentations (HSV, mosaic, mixup), and resolution normalisation for altitude/lighting robustness
- Trained YOLOv8 OBB; conducted ablation studies across resolutions, thresholds, and augmentation configs; reported mAP50, mAP50-95, and P/R curves to faculty supervisors
- Exported model to TensorRT FP16 on NVIDIA Jetson Orin NX; achieved 60–90 FPS GPU inference (~20x CPU baseline speedup)
- Architected a fully offline real-time pipeline: DJI Pilot 2 RTMP → MediaMTX → TensorRT → Flask/WebSocket annotated stream — zero internet dependency
- Built ground station on Jetson acting as WiFi hotspot; RC pushes H264 over RTMP; Jetson serves annotated MJPEG + JSON telemetry to Android tablet over USB-C tethering
- Fused live GPS telemetry (DJI MSDK serial, 10 Hz) per-frame with detections; results timestamped, geotagged, and persisted in SQLite
- Engineered Flutter mobile app with native Kotlin plugin (MediaCodec HW decode, YUV→Bitmap), WebSocket client, offline tile maps, per-detection GPS pins, and SQLite detection log
- Benchmarked TFLite delegates (CPU/GPU/NNAPI) across FP32/INT8; built dual inference backends (TFLite + ONNX Runtime with XNNPACK)
Software Engineer
NUST Flagship Project (Computer Vision — Confidential)
- Curated and cleaned real-world image datasets; designed augmentation pipelines for domain shift robustness
- Fine-tuned YOLOv8 on custom domain-specific object detection task
- Quantised model to float16 and deployed on-device via Flutter for real-time mobile inference
Software Engineer Intern
UpSkillers (Remote)
- Delivered full-stack web features across multiple client projects within an agile remote team
Network Engineer Intern
Quetta Safe City Project
- Optimised network efficiency and security within a large-scale government surveillance infrastructure
Featured Projects
Systems I've designed and shipped — from AI pipelines to scalable full-stack applications.
Space Biology Knowledge Engine
Winner — NASA Space Apps Challenge (BUITEMS, 2025)
AI-powered research platform that transforms NASA space biology papers into a Neo4j knowledge graph and employs a RAG pipeline with vector embeddings for contextual scientific literature retrieval.
Architecture
Healthcare Enterprises Website
Production client website with dynamic product catalogue, MongoDB backend, and polished Next.js/React frontend.
Architecture
Family Tree Software
Feb 2025 – Present
Desktop application supporting 5M+ individuals with graph-traversal search, zoom, and dynamic PDF export.
Architecture
Network Intrusion Detection System
Nov 2024
ML-based NIDS combining Random Forest and LSTM for superior network anomaly detection.
Architecture
Mobile Applications
Cross-platform mobile development with Flutter — from learning platforms to edge AI inference.
CourseUp
Online Learning Platform
Full-stack online learning platform built with Flutter and Node.js. Users can browse, stream, and learn from courses with video playback, category filtering, and RESTful API integration. Backend powered by MongoDB Atlas and Cloudinary for media hosting.
- Course browsing & category filtering
- Video streaming with custom controls
- RESTful API with Express.js
- Cloudinary media hosting
Achievements & Awards
Hackathons, competitions, and recognitions across AI and software engineering.
Gold Medalist — High Achiever Award, NUST (2025)
Awarded for outstanding academic and extracurricular performance.
Winner — NASA Space Apps Challenge (BUITEMS, 2025)
Space Biology Knowledge Engine using AI & NLP for space mission research.
Winner — NUST CS Department Coding Competition
May 2024
Qualified — Regional Round, Ignite Digital Cybersecurity Hackathon
Nov 2024
Certifications
Continuous learning across deep learning, generative AI, and full-stack development.
Deep Learning Specialization — Deeplearning.ai
Generative AI with LLMs — AWS/Coursera
Meta React Basic & Advanced
IBM Node.js & Express
IBM Python for Data Science
AI A-Z 2024 — Udemy
Enrolled: NLP Specialization — Coursera
Get In Touch
Have a project or opportunity? Send a message — I'll get back to you soon.