Covariate enables your company to:- Better estimate your R&D costs & risks, accelerate your product discovery- Identify competition and potential M&A targets- Identify and monitor relevant research, clinical trial & patent activity, which could affect the outcomes of their R&D projects- Identify professionals & emerging talent suited for your open R&D positions or best institutions to collaborate with- Provide infographics for general and professional public, justifying your R&D activities and products http://covariate.com/business/Active

On demand data analytics, AI and machine learning prototypes. Do you have an ML/data related problem that you want to address fast? Give us a call. Is it complicated, novel, uncertain, maybe a bit crazy? Even better.Previous clients: - Kickresume: tool that evaluates quality of resumes. Deployed and running as part of their product offering. - MindMed: Analysis of clinical data in agitation research. Analysis of clinical data for cough detection.

Led cross team project (research ML, applied ML, backend, mobile, crypto) to enable multi-modal biometric authentication.Led research on behavioral authentication and further deployment to production.Led project on data collection for behavioral authentication.Deployed face recognition and identification models to mobile phones. Designed, implemented and maintained pipeline for fully reproducible experiments&research.Active in, hiring, process design, mentoring, internal education.Experience with: computer vision, image processing, signal processing, cross-platform and on the edge ML, behavioral data, facial data, fully remote collaboration and team leading.

Achieved SOTA on CIFAR-100 with AutoML and ensamble learning and large scale architecture search as part of AdaNet team https://arxiv.org/pdf/1903.06236.pdf.Used graph nets for improving branching predictions for mixed integer program solvers and combinatorial optimization in collaboration with DeepMind. Acknowledged in paper https://arxiv.org/pdf/2012.13349.pdf (ML can beat MIP solvers).

Real estate price forecasting- computer vision on top of satellite imageryML researcher focused on NLP problems like:- mention detection- co-reference resolution- relationship extraction- entity centric sentiment analysis

Improved duplicate video detection in core YouTube pipeline for video copyright claims.

Created a tool for visualizing network security incidents https://github.com/log2timeline/eccemotus.Prototyped tool for automatic timestamp recognition.

Assisting CTO at his work. Mostly scripting in Python, data mining, processing, predictions and visualizations. But in general, work of any kind.