Willing to relocate: No, but work with US-based customers all the time (W8-BEN)
Technologies: anything LLM, fine-tuning/alignment, AI agents, custom harnesses; recent Stanford/GT on resume
Résumé/CV: Upon request; I worked on well-known visual AI agents (2024), custom agentic harnesses for a Toptal-like AI agency, automated and accelerated some cognitive sub-processes of a large company, LLM model router etc. My latest project has just concluded and am now looking for a new one after a vacation
Technologies: PyTorch, Huggingface, TensorFlow, Python, Java, JavaScript, AWS, Docker/Kubernetes, Large Language Models, Stable Diffusion, Vision Transformer, OpenCV (tracking, SfM, SLAM etc.), monolith-to-microservice conversion on AWS, AWS cost optimization
CV: Upon request; Deep Learning @ Stanford, ML @ Georgia Tech, business @ UIUC; worked for some well-known companies in SW engineering roles on exciting products before moving to consulting (as an engineering VP)
Technologies: PyTorch, Huggingface, TensorFlow, Python, Java, JavaScript, AWS, Docker/Kubernetes, Large Language Models, Stable Diffusion, Vision Transformer, OpenCV (tracking, SfM, SLAM etc.), monolith-to-microservice conversion on AWS, AWS cost optimization
CV: Upon request; Deep Learning @ Stanford, ML @ Georgia Tech, business @ UIUC; worked for some well-known companies in SW engineering roles on exciting products before moving to consulting (as an engineering VP)
Technologies: Keras/TensorFlow (recent arXiv models), Python, PySpark, Java, C++, CUDA, mostly computer vision & large scale 3D processing and visualization
Résumé/CV:
- manufacturing defect detection for automotive industry with Deep Learning (<2% real-world false positive rate)
- detecting anomalies in human activities from pair-wise spatio-temporal relationships of human pose estimates in continuous video feed using time-distributed attention-enhanced VRNNs
- 3D indoor reconstruction using SLAM, randomized non-linear optimization, semantic segmentation and depth estimation from mobile video feed (ensemble)
- Mobile traffic anomaly/fraud detection (GBT + discrete VRNN)
- Image content filtering using Deep Learning (DenseNet)
- Diagnosing lung diseases from X-Ray images (CheXNet), surpassing human level performance
- End-to-end self-driving car control (NVidia Dave2Net)
- MS from a top US school, worked for some of the best tech companies; detailed CV upon request
Technologies: Keras/TensorFlow (recent arXiv models), Python, PySpark, mostly computer vision & large scale 3D processing and visualization
Résumé/CV:
- manufacturing defect detection for automotive industry with Deep Learning (<2% real-world false positive rate)
- detecting anomalies in human activities from pair-wise spatio-temporal relationships of human pose estimates in continuous video feed using time-distributed attention-enhanced VRNNs
- 3D indoor reconstruction using SLAM, randomized non-linear optimization, semantic segmentation and depth estimation from mobile video feed (ensemble)
- Mobile traffic anomaly/fraud detection (GBT + discrete VRNN)
- Image content filtering using Deep Learning (DenseNet)
- Diagnosing lung diseases from X-Ray images (CheXNet), surpassing human level performance
- End-to-end self-driving car control (NVidia Dave2Net)
- MS from a Top US school, worked for some of the best tech companies; detailed CV upon request
Technologies: Keras/TensorFlow (recent arXiv models), Python, PySpark, mostly computer vision & large scale 3D processing and visualization
Résumé/CV:
- manufacturing defect detection for automotive industry with Deep Learning (<2% real-world false positive rate)
- detecting crime in progress from pair-wise spatio-temporal relationships of human pose estimates in continuous video feed using time-distributed attention-enhanced ConvNet-RNNs
- 3D indoor reconstruction using SLAM, randomized non-linear optimization, semantic segmentation and depth estimation from mobile video feed (ensemble)
- Mobile traffic anomaly/fraud detection (GBT + discrete VRNN)
- Image content filtering using Deep Learning (DenseNet)
- Diagnosing lung diseases from X-Ray images (CheXNet), surpassing human level performance
- End-to-end self-driving car control (NVidia Dave2Net)
- Top 10 US school MS education; detailed CV upon request
Technologies: Keras/TensorFlow (recent arXiv models), Python, PySpark, mostly computer vision & large scale 3D processing and visualization
Résumé/CV:
- manufacturing defect detection for automotive industry with Deep Learning (<2% real-world false positive rate)
- detecting crime in progress from pair-wise spatio-temporal relationships of human pose estimates in continuous video feed using time-distributed attention-enhanced ConvNet-RNNs
- 3D indoor reconstruction using SLAM, randomized non-linear optimization, semantic segmentation and depth estimation from mobile video feed (ensemble)
- Mobile traffic anomaly/fraud detection (GBT + discrete VRNN)
- Image content filtering using Deep Learning (DenseNet)
- Diagnosing lung diseases from X-Ray images (CheXNet), surpassing human level performance
- End-to-end self-driving car control (NVidia Dave2Net)
- Top 10 US school MS education; detailed CV upon request
Remote: Yes
Willing to relocate: No, but work with US-based customers all the time (W8-BEN)
Technologies: anything LLM, fine-tuning/alignment, AI agents, custom harnesses; recent Stanford/GT on resume
Résumé/CV: Upon request; I worked on well-known visual AI agents (2024), custom agentic harnesses for a Toptal-like AI agency, automated and accelerated some cognitive sub-processes of a large company, LLM model router etc. My latest project has just concluded and am now looking for a new one after a vacation
Email: deepmodel at protonmail.com