Abdullah Ahmed Nawaz

Abdullah Ahmed Nawaz

AI/ML Engineer specializing in Agentic AI and Multi-Agent Systems

Based in: Karachi, PakistanMost recently: AI/ML Engineer (Intern), Chrio

Hire me

Artificial Intelligence graduate with extensive experience building LLM-powered applications, RAG pipelines, and multi-agent systems using LangChain and LangGraph. Proven track record in developing end-to-end AI solutions, from computer vision simulations to federated MLOps health risk prediction systems.

Experience

AI/ML Engineer (Intern)

ChriointernshipMay 2025 – Aug 2025

Summary

Worked in the R&D team on a computer vision-based batsman simulation system using real match footage. Contributed to research, visual data preparation, annotation of 3,000+ images, feature engineering, and model evaluation.

Results

  • Annotated 3,000+ images for visual data preparation.
Computer VisionFeature EngineeringModel Evaluation

Skills

Technical

PythonPython
LLM Applications
LangChain
JSON
LangGraph
Agentic AI
FastAPIFastAPI
Tool Calling
RAG
Prompt Engineering
Embeddings
Semantic Search
NumPy
Pandas
OpenCVOpenCV
GitGit
DockerDocker
Structured Outputs
Streamlit
Scikit-learn
REST APIs
SQLSQL
PyTorchPyTorch
TensorFlowTensorFlow
WebSockets
C++C++
JavaScriptJavaScript
HTML/CSSHTML/CSS
GitHub ActionsGitHub Actions
JenkinsJenkins
Node.jsNode.js
ReactReact
Adversarial Evaluation
gRPC
Parler TTS
Structured JSON prompting
Response validation
Ollama

Areas of expertise

Explaining concepts simply
Teamwork
Presentation skills
Clear communication

Projects

MCVT: Multi-Camera Vehicle Tracking and 3D Ground-Plane Visualization

Lead Developeracademic
Built a real-time multi-camera vehicle tracking system to assign consistent Global IDs across five synchronized camera feeds. Implemented IoU-based ground-plane footprint matching, vehicle-class filtering, and clustering for cross-camera re-identification. Developed a WebSocket backend, camera-stream emulator, and interactive web dashboard for live tracking and camera journey monitoring. Integrated Unity WebGL 3D visualization to display vehicle trajectories and ground-plane movement across camera views.
WebSocketUnity WebGL3D VisualizationPython

AutoServe AI, Multi-Agent Startup Automation System

personal
Built a Python-based multi-agent system with CEO, Product, Engineer, Marketing, and QA agents communicating through a structured JSON message bus. Automated landing-page generation plus real GitHub issue/branch/commit/PR creation, Slack bot updates, and email delivery via Gmail SMTP using OpenAI-compatible LLMs and QA-driven revision loops.
PythonJSONGitHub APISlack APIGmail SMTPOpenAI LLMs

Trust-Aware Multi-Agent Travel Planner

personal
Built a multi-agent system with Planner, Tool, Trust Governor, and Verifier agents using shared memory, verification logic, deterministic tools, and LangGraph-based execution. Designed clean and adversarial evaluation scenarios to measure agent reliability, contamination, recovery, and failure behaviour
LangGraphPythonShared MemoryOllama

Story2Audio, NLP

personal
Built a gRPC-based microservice for text-to-speech using the Parler-TTS Mini Jenny 30H model, integrating Hugging Face AutoTokenizer for preprocessing and handling concurrent audio generation requests via a server-side validation pipeline. Implemented end-to-end pipeline with Postman-tested gRPC APIs, input validation (length, characters), and CPU-based inference, returning WAV audio byte streams with performance evaluated under concurrent load.
gRPCParler-TTSHugging FaceAutoTokenizerPostmanPython

EnviroSense 360, Federated MLOps Health Risk Prediction System

personal
Developed a federated learning-based health risk prediction system using manual FedAvg across distributed clients, integrating multi-source data (WESAD + air quality) with preprocessing, drift detection, and conditional retraining pipelines. Deployed production-grade inference using FastAPI with Docker containerization and CI/CD (GitHub Actions), enabling real-time predictions, logging, and dual Streamlit dashboards for citizen and authority-level monitoring.
Federated LearningFedAvgFastAPIDockerGitHub ActionsStreamlitMLOps

Education

Bachelor of Science (BS)

FAST-NUCES - National University of Computer and Emerging SciencesArtificial IntelligenceMay 2026

Thesis. MCVT: Multi-Camera Vehicle Tracking and 3D Ground-Plane Visualization

Hire AbdullahReach out about a role, a contract or a conversation.For recruiters

Abdullah's twin is AI, it can make mistakes.

More people on Kanvis

Browse more peopleCreate your own Kanvis