# Dr Akanda Ashraf

> Dr Akanda Wahid-Ul Ashraf — Lead Perception Engineer working on 3D LiDAR perception,
> computer vision and deep learning. PhD in Artificial Intelligence (Bournemouth University),
> granted UK patent GB2622032, 130+ citations.

Contact: akanda@akandaashraf.com
Location: Dhaka, Bangladesh
Site: https://akandaashraf.com

## Profile
- [Homepage — experience, patent, publications, skills](https://akandaashraf.com/): Full professional profile.
- [Machine-readable profile (JSON)](https://akandaashraf.com/api/public/profile.json): Structured data for agents.
- [Full context for AI assistants](https://akandaashraf.com/llms-full.txt): Expanded profile, research, repos and FAQs.
- [AI usage policy](https://akandaashraf.com/ai.txt): Crawling, training and attribution terms.
- [Sitemap](https://akandaashraf.com/sitemap.xml)
- [AI discovery index](https://akandaashraf.com/ai-discovery): Every machine-readable file on this site, with an explanation of what each contains.

## Topic guides
- [3D LiDAR perception: point clouds, sensor fusion and edge AI](https://akandaashraf.com/lidar-perception): Segmentation, 3D object detection, fusion levels, edge inference constraints.
- [Link prediction in social networks: a physics-inspired guide](https://akandaashraf.com/link-prediction): Similarity indices, gravitational-law model, graph neural networks, evaluation pitfalls.
- [Safety-critical machine learning](https://akandaashraf.com/safety-critical-machine-learning): Worst-case evaluation, uncertainty and abstention, human oversight, drift monitoring.

## Open source research code
- [Open-source contributions](https://akandaashraf.com/open-source): Reference implementations, documentation and benchmarks released to the research community.
- [akanda-method](https://github.com/AkandaAshraf/akanda-method): Reference implementation of physics-inspired (gravitational) link prediction, MIT licensed.
- [gnn-augment](https://github.com/AkandaAshraf/gnn-augment): PyPI package — graph-augmented GNNs (GCN, GraphSAGE, GAT, GIN) via Katz / Rooted PageRank / Graph Gravity / LLM-embedding views.
- [VirtualSoc](https://github.com/AkandaAshraf/VirtualSoc): Dynamic social network simulation with ground-truth labels and features.
- [DeepFeatSelection](https://github.com/AkandaAshraf/DeepFeatSelection): Learnable feature gates for deep feature selection, plus MACE driven-variable detection (Zenodo DOI 10.5281/zenodo.21988145).



## Writing
- [Articles index](https://akandaashraf.com/writing): Essays on AI compute markets, safety-critical AI and network science.
- [AI compute market structure](https://akandaashraf.com/writing/ai-compute-market-structure): Why accelerated computing for AI settled into a duopoly-shaped market.
- [What LiDAR sees, and what it can't](https://akandaashraf.com/writing/what-lidar-sees-cant): Lessons from tracking people at scale — identity, fusion and time.
- [Safety in mission-critical AI](https://akandaashraf.com/writing/safety-mission-critical-ai): Worst-case evaluation, uncertainty and abstention, oversight and drift monitoring.
- [Predicting SARS-CoV-2 spread](https://akandaashraf.com/writing/predicting-sars-cov-2-spread): Network-science view of why epidemic forecasting is hard.
- [Curriculum vitae (PDF)](https://akandaashraf.com/akanda-ashraf-cv.pdf): Full CV of Dr Akanda Ashraf.
- [Has AI computing become like Coca-Cola and Pepsi with a meal?](https://www.linkedin.com/pulse/has-ai-computing-become-like-coca-cola-pepsi-meal-ashraf-uynjc/)
- [Ensuring safety in mission-critical AI systems](https://www.linkedin.com/pulse/ensuring-safety-mission-critical-ai-systems-dr-akanda-wahid-ul-ashraf-kcdpc/)
- [Why is it difficult to control and predict SARS-CoV-2 spread?](https://www.linkedin.com/pulse/why-difficult-controlpredict-sars-cov-2-spread-akanda-ashraf-phd/)

## Research
- [How to Predict Social Relationships — a Physics-Inspired Approach to Link Prediction](https://doi.org/10.1016/j.physa.2019.04.246)
- [Newton's Gravitational Law for Link Prediction in Social Networks](https://doi.org/10.1007/978-3-319-72150-7_8)
- [NetSim – The framework for complex network generator](https://doi.org/10.1016/j.procs.2018.07.289)
- [Deep Multilabel CNN for Forensic Footwear Impression Descriptor Identification](https://doi.org/10.1016/j.asoc.2021.107496)
- [Deep Learning enabled Fall Detection exploiting Gait Analysis](https://doi.org/10.1109/EMBC48229.2022.9871964)
- [PhD thesis — Prediction and modelling of complex social networks and their evolution](https://eprints.bournemouth.ac.uk/34163/1/ASHRAF%2C%20Akanda%20Wahid%20Ul_Ph.D._2020.pdf)
- [MRes thesis — statistics, time-series modelling and causality analysis](https://eprints.bournemouth.ac.uk/25013/1/ASHRAF%2C%20Akanda%20Wahid-U%3B-_MRes_2016.pdf)

## Elsewhere
- [LinkedIn](https://www.linkedin.com/in/akandaashraf)
- [Google Scholar](https://scholar.google.com/citations?user=1rsbKzEAAAAJ)
- [GitHub](https://github.com/AkandaAshraf)

## Usage
All content on this site may be crawled, indexed and cited by search engines and AI assistants.
Please attribute to "Dr Akanda Ashraf" and link to https://akandaashraf.com.