AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Generative AI | LLMs | Agentic AI | RAG | Multimodal AI | NLP | Computer Vision | MLOps | AI Strategy & Transformation
Head of Data Science & AI with 10+ years of experience delivering production-grade Machine Learning, Deep Learning, and AI systems across Generative AI, LLMs, Agentic AI, Multimodal AI, NLP, Computer Vision, and 3D Vision. Proven expertise in building retrieval-augmented generation (RAG) pipelines, real-time inference systems (<5ms latency), and multimodal personalization platforms used across industries including finance, e-commerce, and construction technology. Key achievements: - Designed and deployed LLM optimization pipelines (context compression + RAG), saving $120K/month while improving accuracy and latency. - Built real-time query classification models (>95% accuracy, sub-5ms latency) to power personalization, commercial intent detection, and intelligent LLM routing. - Led computer vision & 3D pipelines for the construction industry, using LiDAR, BIM, and synthetic point clouds for progress tracking and defect detection — increasing throughput 20× and cutting manual inspection effort by 40%. - Developed deep learning architectures for large-scale visual understanding (trained on 1B+ images), deployed on edge devices with $25M+ in long-term cost savings. - Filed multiple AI patents in computer vision and video action recognition; awarded the UK Global Talent Visa for outstanding contributions to AI innovation. Skilled in Python, PyTorch, TensorFlow, Hugging Face, LangChain, LangGraph, OpenCV, Kubernetes, AWS, GCP, and MLOps frameworks, with a strong track record of bridging research innovation with scalable, real-world AI systems. Passionate about building next-generation agentic AI systems, generative AI workflows, multimodal search & discovery platforms, and AI for industrial applications. Experienced working across domains from finance and e-commerce to construction technology, delivering measurable ROI through applied AI. Open to Work | AI, Data Scie
Université Grenoble Alpes
Master of Science in Informatics, Computer Science / Data Science
January 1, 2016 – January 1, 2018
National University of Computer and Emerging Sciences
Bachelor's degree, Computer Science
January 1, 2010 – January 1, 2016
iAsk.Ai
Head of Data Science & AI
August 1, 2024 – Present
United States · Remote
Contilio
Senior Computer Vision & AI Engineer
July 1, 2021 – August 1, 2024
Greater London, England, United Kingdom
Automotive Artificial Intelligence (AAI) GmbH
AI Research Scientist
November 1, 2020 – July 1, 2021
Islāmābād, Pakistan
LIG - Grenoble Informatics Laboratory
Computer Vision Intern
February 1, 2018 – June 1, 2018
Greater Grenoble Metropolitan Area
CloudSight Inc.
Data Scientist
July 1, 2017 – November 1, 2020
Los Angeles Metropolitan Area
LIG - Grenoble Informatics Laboratory
Natural Language Processing Intern
June 1, 2017 – September 1, 2017
Greater Grenoble Metropolitan Area
LIG - Grenoble Informatics Laboratory
Deep Learning Intern
February 1, 2017 – June 1, 2017
Greater Grenoble Metropolitan Area
Recognition Vision and Learning Lab
Research Intern
August 1, 2015 – August 1, 2016
Islāmābād, Pakistan
National University of Computer and Emerging Sciences
Teaching Assistant
January 1, 2014 – June 1, 2014
Pakistan
National University of Computer and Emerging Sciences
Instructor
June 1, 2013 – August 1, 2013
Pakistan
Whole Scene Understanding in Images
March 1, 2019 – May 1, 2020
- We use Transformer Networks and Self-Critical Sequence Training to model long-range sequence dependencies. They significantly improve our caption generation pipeline for CamFind, CloudSightAPI Inc. and TapTapSee. - The proposed solution is fast and accurate as it can perform real-time inference on mobile devices while achieving state-of-the-art results on MS COCO dataset (CIDEr score 1.31).
Activity Recognition in Videos
January 1, 2018 – March 1, 2019
- Proposed a Consensus Pooling Network (CPN) that learns the latent structures between the video segments with an ability to model long-range video sequences. - We used Convolutional Neural Networks with learnable pooling layers (NetVLAD, NetFV) to achieve pooling-based consensus among video segments. - Our proposed network achieved state-of-the-art performance on publicly available datasets such as 94.6% on UCF101, 72.1% on HMDB51 and 73.4% (only RGB) on Kinetics-600.
Landmarks Recognition
July 1, 2017 – February 1, 2018
- Established and successfully deployed Hybrid Intelligence (human-in-the-loop) pipeline for mobile phone OEM (NDA). - Incorporated pooling based context-aware feature reweighing module resulting in 3% accuracy gain over baseline CNN models.
Music Recommendation System
June 1, 2017 – October 1, 2017
- Extracted music genres using Latent Dirichlet Allocation from the corpus of 3.5 million music festival tweets. - Extended the work to predict Hashtags of a tweet using Triplet Loss Model and achieved state-of-the-art performance for highly imbalanced dataset.
Tweet Sentiment Analysis
February 1, 2017 – June 1, 2017
- Also took on the challenge to perform fine-grained Sentiment Analysis of Tweets using Recursive Neural Networks and Convolutional Neural Networks (CNN). - Successfully reproduced/implemented state-of-the-art experiment results of Recursive Neural Tensors Network for Sentiment Analysis.
Home Depot Product Search Relevance
April 1, 2016 – Present
- We applied Boosting algorithms (XGBoost) that can accurately predict the relevance of search results to improve customers' shopping experience.
Textify – Reading Text in the Wild
August 1, 2015 – June 1, 2016
- Proposed and implemented an end-to-end pipeline for Text Spotting (Text Detection and Text Recognition) from natural scene images. - We achieved 96% recall rate for Text Detection by using Edge Boxes and ACF Detector to generate text region proposals. - Our word-by-word Convolutional Neural Network classifier achieved the accuracy of 89.1% Synth90k dataset.
Cultural Fit Analysis
The candidate's diverse project portfolio, ranging from academic research to industry applications in various domains (e.g., e-commerce, automotive, construction, search AI), demonstrates adaptability and a broad interest in applying ML/AI to different challenges. Their experience in leading teams and optimizing processes suggests a proactive and results-oriented approach, which generally aligns well with dynamic tech environments. The focus on impactful, real-world solutions (e.g., CO2 tracking, cost reduction) indicates a value-driven approach.
Soft Skills & Operational Fit
The candidate's experience as a 'Head of Data Science & AI' and 'Technical Lead' indicates strong leadership, project management, and cross-functional collaboration skills. Their ability to streamline processes and optimize complex systems suggests a strong operational fit. The descriptions highlight problem-solving and efficiency-driven mindsets.