Kuwait City — Software Engineer & Technical Lead
I build products.
Software engineer with an M.S. from Michigan State and a founder’s job description: I architect platforms end to end, turn inherited codebases into production systems, and train everyone from engineers to educators along the way.
I’m the engineer teams hand things to when they need them shipped. At Novagates I inherited a struggling platform from a previous team, led the disaster recovery — about 80% of the critical issues resolved — rebuilt the engineering team from scratch, set up the GitHub organization and deployment pipelines, and took the product to live production. Now I lead three distributed teams there while still writing code every week, working directly with the founders, board members, and senior engineers on where the platform goes next.
I also run Akasi, the startup I founded — an AI assistant that turns basic product photos into brand-consistent marketing visuals for small businesses. Every layer of it is mine, from the GCP infrastructure and Django backend to the Next.js frontend and the prompt engineering behind the image generation.
The third thread is teaching: through the Google AI initiative with Kuwait’s Ministry of Education, I train supervisors and teachers to use AI well — its limits and failure modes included. Before all this came an M.S. in Computer Science at Michigan State (GPA 3.95) with research in ML security, and a B.S. with High Distinction from the University of Michigan-Dearborn.
Kuwait City · open to collaboration
remote teams led at Novagates
of inherited critical issues resolved
graduate research projects
M.S. GPA at Michigan State
Founder, technical lead, and engineer — usually all three in the same week.
Jun 2025 — Present · Remote
Mar 2026 — Present · Kuwait City · Part-time, Hybrid
Novagates
Apr 2026 — Present · Kuwait City · Full-time, Hybrid
Bambr
Sep 2025 — Apr 2026 · Kuwait
MTS — Ministry of Health
2024 — 2025
Michigan State University · East Lansing, MI
GPA 3.95 · Focus on machine learning, ML security, and distributed systems.
2019 — 2023
University of Michigan-Dearborn · Dearborn, MI
High Distinction · GPA 3.84 · Honorable Mention, Senior Design Day 2023.
Systems I designed, built, and shipped — from a live AI product to an award-winning client platform.
A live product that turns a basic product photo into polished, on-brand marketing visuals. Built, deployed, and marketed end to end as a solo founder: micro-monolithic backend, GCP infrastructure with a secured CDN, tiered background tasks, multi-vendor payments, and prompt-engineered image generation.
Interactive brain-tumor segmentation
A medical-imaging platform that detects and segments brain tumors from MRI scans using U-Net and DeepLabV3 trained on BraTS 2020, with DCGAN-based augmentation and CLIP-enhanced variants. An interactive Next.js + FastAPI viewer renders segmented tumors in 3D in real time.
Full-stack app shipped to the App Store
A production application built for a real client: requirements gathering, system analysis, a Django + NGINX backend, and a React Native app I led through App Store deployment. Recognized with an Honorable Mention at UM-Dearborn’s Senior Design Day competition.
Honorable Mention · Senior Design Day 2023
Good engineering deserves good explanation. I design and deliver AI training for ministries, institutes, and professional audiences.
Bambr · Kuwait · 2026 — ongoing
Ongoing AI training delivery across Kuwait’s Ministry of Education: hands-on sessions that get supervisors, head teachers, and classroom teachers actually using Google AI products in their daily work — grounded in responsible use, not hype.
KGIC · 3-hour workshop · Jan 2026
How and when to use modern AI at work: why models can be confidently wrong, concrete mitigation strategies, how these systems actually work under the hood — simplified for a non-technical audience — and prompt engineering fundamentals.
KGIC · 3-hour workshop · Oct 2025
A practical walkthrough of role-specific AI applications and how to fold them into daily workflows to boost productivity and creativity in professional environments.
Google Certified Educator — Level 1 & Level 2
Hands-on graduate research from my M.S. at Michigan State — ML security, applied deep learning, and optimization. Full write-ups available as PDFs.
Michigan State University · Apr 2025
A. Houtari, S. Shakir, L. Mo
98.77% accuracy · 91% recall (best single model)
A defensive layer for LLMs that flags harmful prompts without expensive retraining. An autoencoder trained only on harmful prompts treats benign text as out-of-distribution, and a weighted-vote ensemble over sentence-transformer embeddings adds robustness against adversarial examples.
Michigan State University · May 2025
A. Houtari, D. Mohanty
Implements BadNets on CIFAR-10 and the Hidden Trigger Backdoor Attack on TinyImageNet, then probes whether Neural Cleanse can reverse-engineer both. The supposedly stealthy attack still left detectable traces — with novel anomaly-detection metrics beyond the original protocol.
Michigan State University · Dec 2024
A. Houtari, D. Mohanty, A. Lochab, S. Shakir, S. Park
Brain-tumor classification and segmentation on BraTS 2020, using DCGANs for data augmentation and comparing U-Net, DeepLabV3, and CLIP-enhanced variants with Dice and IoU metrics. Paired with an interactive 3D visualization platform.
Michigan State University · Dec 2024
A. Houtari, A. Lochab, S. Shakir, S. Situ
Detects political bias in Western media using a CNN-GRU model with GloVe embeddings on the MBIC and NewB datasets, benchmarked against transformer baselines like BERT and GPT-4 across liberal, neutral, and conservative labels.
Michigan State University · Dec 2024
A. Houtari, S. Shakir
Tackles the NP-hard Vehicle Routing Problem with genetic algorithms, comparing single- and multi-objective fitness functions (including NSGA-II) and 2-opt local search under capacity constraints.
Michigan State University · May 2024
A. Houtari, D. Mohanty, K. Suzue
Extends ant-colony search algorithms for peer-to-peer networks with LSTMs that tune pheromone and TTL parameters in real time from historical data, improving search reliability and relevance over AntP2PR and DLAntP2P. Simulated in PeerSIM.
Also: re-implemented SSPCAB (CVPR 2022) — self-supervised anomaly detection for computer vision — in both TensorFlow and PyTorch. View on GitHub
My capstone team earned an Honorable Mention at the Senior Design Day competition held by the College of Engineering and Computer Science at the University of Michigan-Dearborn — recognized among many strong projects for creativity, technical execution, and real-world applicability.
Google Certified Educator — Level 1 & Level 2
07 — Contact
A product idea, a team that needs a technical lead, an AI training program — my inbox is open. I read everything and reply to anything interesting.