Michael Misiewicz Email & Phone Number

New York, Ny, Us

New York, Ny, Us

I lead all ML, AI and Data Science efforts at Yext. The department's work improves Yext's products: Search, Chat, Reviews, Listings, and Knowledge Graph. We are heavily invested in large language models.I joined in 2019 to build the department and boot up a data science practice. I'm responsible for:- managing hiring, performance, budgets and career architecture for the organization. - Setting strategy, vision, roadmaps and research objectives - Managing managers- Partnership, education and collaboration with ML Engineering, Executives and the BoardSome of my major accomplishments:- Deployed Large Language Models in production (GPT-NeoX, BloomZ and InstructGPT).- Developed powerful data mining tools (DBSCAN and HDBCAN powered clustering models). Built automatic cluster summarization and anomaly detection algorithm.- Built the production model serving infrastructure (Python in Docker containers)- Built a robust data governance and labeling organization under my supervision - including hiring the leader of Labeling Operations (all good data science projects start with strategic data assets!)- Built model training pipelines and batch prediction infrastructure. Also developed initial data warehouses.- Shipped models for traditional ML problems (e.g. Listings Matching)- Delivered fundamental research on Document Search, finding an optimal HNSW and SentenceTransformers powered system, allowing a total rebuild of one of Yext's search backends.- Delivered research and models for boilerplate removal from web crawler output.- Long-term research program to develop an GPT-powered pipeline to extract knowledge graphs from unstructured documents- Working with marketing, UX and the Web Team, we instrumented all of Yext's web presence to build churn prediction models, recommender algorithms, and personalization algorithms to increase customer adoption of Yext productsAll of this will be coming to the Yext releases over the next 12 month period!

New York, Ny, Us

Building a search engine! Some areas I'm currently working on:- Bayesian methods for NLP problems- Deep neural networks for entity recognition and extraction (BERT and other attention architectures in particular) - Production implementations and QA of the above- Data warehouse architecture and implementation - A/B testing and experimentation

Working on bespoke analysis and problem solving for select clients. My projects have been focused on brining cutting edge methods and know-how in a cost effective manner to clients who need the hardest problems solved in the shortest timeframe.

New York, New York, Us

I managed a team of 5 data scientists responsible for evaluating and prototyping algorithms for detection of fraudulent content (e.g. "fake news" sites) and users (botnets) in AppNexus's systems. This means processing big data using unsupervised machine learning on 350 billion events per day (in streaming and in batch) and also supervised Natural Language Processing to analyze and interpret content around the web. I worked closely with product managers, executives and engineering leaders to prototype, plan and implement detection systems. Growth and development of everyone on my team was of the utmost importance to me, and one of my most important responsibilities. On a daily basis, my team and I worked in:- Big data tools (Spark, Postgres, Python [pandas, matplotlib], Scala, Airflow, Presto)- Machine learning software (nltk, scikit-learn, NetworkX)- Dimensionality reduction (t-SNE, SVD, PCA)- Supervised classification (Random Forests, Gradient Boosted Trees, Neural Networks, Regularized regressions, time series analysis)- Unsupervised classification and information retrieval (feature generation, data normalization, DBSCAN, K-means, graph based approaches, MinHash, information theory)- Statistical Approaches for monitoring of tradeoffs between true and false positives, descriptive statistics.

New York, New York, Us

My responsibility is to lead the design of strategy, roadmaps, techniques and execution of AppNexus's approximately 30 algorithmic anti-Non Human traffic systems developed by the Signals Intelligence team. Our research program includes features from every part of the AppNexus platform and substantial forensic research on viruses. I lead a group of 4 data scientists.

New York, New York, Us

I am researching fraud on the AppNexus platform as a member of our new Signals Intelligence Group. My responsibility is to prototype, validate and experiment with systems to stop advertising fraud and malicious activity on the AppNexus platform. It's a big, fascinating problem! My prototypes and research are used by C-level executives and product managers at AppNexus. I am also working very hard on vetting external partners and vendors for their fraud data sets and solutions. I am working on internet and computer security problems, using the following tools and packages:- Spark, Spark Streaming and Kafka to process streams with over 1.5 MM records / second. I triage, cluster, classify and learn from AppNexus's enormous data sets.- Forensic computing packages: VirtualBox, Wireshark, iptables, tcpdump and other unix hacking tools- Pig, Hive, Parquet and the rest of the Hadoop ecosystem- Python, R, Pandas, other scientific computing packages- Postgres, MySQL for very large data sets

New York, New York, Us

Using scientific approaches and big data tools like Hive and Spark, I conducted research and investigations into clients' Real Time Bidding campaigns on the AppNexus platform. I presented findings and strategy recommendations to executives.

Studied the mechanisms of Endoplasmic Reticulum stress mediated Prion upregulation in human neurons, using bioinformatics techniques. Responsible for experimental design, result processing (in Python and R), and conducting experiments in the lab. Primarily I worked with PCR, qPCR, Western Blots, Cell Culture and Luciferase assays.

- Researched the effects of Prion diseases (i.e. mad cow disease or Creutzfeldt-Jakob Disease) on the ability of human cells to undergo programmed cell death. - Researched the interaction of the kinase CDK5 and the Prion protein using radioactive tracers.

Chasseneuil-De-Poitou, Fr

Developed English curriculum and lesson plans for 6-12 year-old students in the Paris public school system.

Master Of Science (M.Sc.), Cell Biology And Computer Science

Bsc, Biology