AI Engineer · Edge AI & Computer Vision Specialist

Malik Anees Ahmed

I'm an AI engineer in Islamabad who builds computer-vision systems that run on the device — no cloud, no lag. From 30 FPS face-mesh inference to 5–20 ms vector search, I care about AI that works in the real world: offline, and under pressure. Right now I'm leading Aegis Drive, my final-year project at NUML, and teaching myself RAG and n8n automation along the way.

BS Artificial Intelligence · NUML · Elected Class Representative

⬇ Download Resume Email Me GitHub LinkedIn
  • 30 FPS On-device inference Aegis Drive
  • 5–20 ms Vector search p99 Event ReID Engine
  • −40% False-positive alerts Aegis Drive
  • 100% Post-drive data retention Aegis Drive
Portrait of Malik Anees Ahmed, AI Engineer based in Islamabad
Islamabad, PK Open to AI / Computer Vision roles

Profile

Architecting high-performance AI systems for the edge and resource-constrained environments.

Professional Summary

AI Engineer specialising in end-to-end ML systems, real-time computer vision, and on-device model optimisation with TensorFlow Lite. Delivered 30 FPS face-mesh inference on Android, 5–20 ms Qdrant HNSW vector search in production, and a 40% reduction in false-positive drowsiness alerts through dynamic EAR thresholding.

Experienced in FastAPI (ASGI) backend development, vector-database management (Qdrant), and offline-first mobile AI pipelines deployed on resource-constrained edge devices — now extending into RAG pipelines and workflow automation. Elected Class Representative, BS AI batch — NUML Islamabad.

Education

BS in Artificial Intelligence National University of Modern Languages (NUML), Islamabad 2023 – 2027 (Expected) ✓ Elected Class Representative, BS AI batch
Intermediate in Computer Science (ICS) Punjab Group of Colleges, Blue Area Campus, Islamabad 2020 – 2022

Key Competencies

  • End-to-end ML pipeline design — data ingestion through TFLite / ONNX model deployment on-device
  • Real-time computer vision on Android using CameraX + MediaPipe at 30 FPS with a 478-point 3D face mesh
  • Offline vector-search microservices with FastAPI + Qdrant HNSW at 5–20 ms p99 latency
  • Person Re-Identification using dual-vector Reciprocal Rank Fusion — 512-D ArcFace + 768-D DINOv2 ViT with 2.5× facial weighting
  • Edge AI driver-safety systems — Temporal BiLSTM TFLite, dynamic multi-tiered EAR / MAR drowsiness detection
  • RAG pipeline design; YOLO11 object detection; Scikit-learn classical ML
  • Offline-first navigation with MapLibre GL JS + OSRM + MBTiles tile serving, zero cloud dependency
  • Production-grade FastAPI (ASGI) backends with atomic Firebase Firestore telemetry — 100% data retention
  • MVVM / Clean Architecture on Android (Kotlin JDK 17), Material Design 3, DND-bypass alarm systems
  • Windows file-lock race-condition diagnosis and median-based embedding consistency checks in Python

Experience & Leadership

Building software in teams — and being trusted to represent them.

2023 – Present Leadership

Class Representative — BS Artificial Intelligence

National University of Modern Languages (NUML), Islamabad

  • Elected by the BS AI cohort to represent peers and act as the liaison between students and faculty.
  • Coordinate academic matters, scheduling, and feedback across the batch throughout the degree.
Mar 2022 – Sep 2022 Role

Front-End Web Developer

NAVTTC Certification Program — Dynamic Web Applications & Interfaces

  • Developed and deployed mobile-responsive web applications using HTML5, CSS, and Firebase Hosting.
  • Completed the certification programme with an A+ grade — the highest achievable in the cohort.
  • Built production-ready frontend interfaces focused on mobile-first layout, performance, and accessibility.
NAVTTC Certified — Grade A+

How I Work

From raw signal to a model running on the device.

  1. Data & Signal

    Capture and clean real-world CV data at the source — CameraX frame streams, MediaPipe 478-point landmarking, and embedding generation with ArcFace and DINOv2.

    CameraX MediaPipe OpenCV

  2. Model & Optimise

    Train and fine-tune in PyTorch, then quantise and convert to TFLite / ONNX for the target device — trading accuracy against latency until it fits the hardware budget.

    PyTorch TFLite ONNX

  3. Deploy to the Edge

    Ship offline-first: TFLite on Android, FastAPI + Qdrant microservices, and MapLibre / OSRM navigation — engineered to run with zero cloud dependency.

    FastAPI Qdrant Android

  4. Monitor & Harden

    Close the loop with atomic Firestore telemetry (100% retention), dynamic thresholding to cut false positives, and race-condition & consistency checks in production.

    Firestore Telemetry EAR tuning

Currently learning Extending this workflow into n8n workflow automation and agentic RAG — automating the glue between data, models, and delivery.

Technical Matrix

The tools I reach for — filter by discipline.

Core Projects

Three systems, three architectures — each shipped end to end.

Certifications

Verified credentials and continuous learning.

NAVTTC Front-End Web Developer Certification, Grade A+, issued January 2023

NAVTTC · Jan 2023

Front-End Web Developer Certification

Certificate in IT – Web & Mobile Web Development. Completed with Grade A+ through the National Vocational & Technical Training Commission programme.

DeepLearning.AI · Jul 2025

AI For Everyone

Completed Andrew Ng's "AI For Everyone" on Coursera — foundational coverage of AI strategy, terminology, and practical deployment across technical and non-technical audiences.

Issued to “Malik Ahmed” (as printed) · verify code 2V1H0DEVQN43

Open in New Tab ↗

Contact

Available for AI engineering, edge ML, and computer vision work.

Get In Touch

Location Gulberg Green, Islamabad, Pakistan
Email malikaneesahmed530@gmail.com
Phone +92-318-5142566

Online Profiles

LinkedIn linkedin.com/in/malik-anees-ahmed
GitHub github.com/MalikAnees530
Focus FastAPI · Qdrant · TFLite · Android · CV · RAG

Academic References

Dr. Moiz Ullah Ghouri Assistant Professor, Dept. of Computer Science, NUML Islamabad mghouri@numl.edu.pk +92-333-5185049
Dr. Farnaz Akbar Assistant Professor, Dept. of Computer Science, NUML Islamabad frnaz.akbar@numl.edu.pk +92-334-5522679
Start a Conversation Browse My Code ↗ ⬇ Download Resume