Martin Tabikh

- Develop and deploy custom AI solutions to help businesses operate smarter and innovate faster. - Deliver tailored AI services that integrate directly into client workflows for measurable impact. - Ensure seamless AI adoption by building solutions that demonstrably boost productivity and create real business value.

- MLOps/LLMOps Lead for the Secure GPT project, an internal conversational assistant based on generative AI that facilitates secure employee interaction with their professional documents. - Design and implementation of preprocessing, ingestion, inference, and evaluation pipelines for language models.

GenAI-Based Application Development: - Virtual Scrum Assistant: Automates daily stand-ups, updates Jira tasks, and proactively identifies and resolves blockers to improve workflow efficiency. - Intelligent Payment Reminder System: Analyzes emails, detects payment-related messages, and automatically updates invoice statuses, reducing manual follow-ups. - RAG-Powered Search Engine for RFPs: Helps retrieve relevant information to efficiently respond to new requests. - Excel File Analysis Tool: Assists in interpreting financial statements, identifying key trends, and supporting data-driven decision-making.

- Development of an intelligent classifier to categorize system failures. - Use of NLP techniques to classify labels and flags from texts, leveraging sentence embeddings derived from transformers.

- Collection of customer reviews from websites in 9 different countries, followed by the application of sentiment analysis and classification algorithms to identify positive points and areas for improvement for each product. - Code optimization, reducing execution time by half. - Implementation of data ingestion, training, and inference pipelines using Vertex AI, DBT, and BigQuery. - Creation of CI/CD pipelines, deployment of dozens of ML models to production, and management of artifacts (libraries and models) as well as the deployment strategy.

- Optimized code for ingesting data from 73 providers on a weekly/daily basis. - Developed a distributed algorithm using Databricks and PySpark to correct over 43% of brand, supplier, and product names. - Developed and maintained a Streamlit web application hosted on Azure to store additional manual corrections of product information.

- Implemented CI/CD pipelines with Azure, managing deployments and automatically generating release notes. - Developed a Streamlit web application and an API with multiple endpoints triggered daily. - Managed configurations and connections with various Snowflake databases. - Optimized code, reducing execution time by 80%.

- Developed Python libraries for classification models on text and tabular data. - Developed hyperparameter optimization models. - Implemented CI/CD pipelines.

Mentor for the following career paths: - Data Scientist - Machine Learning Engineer - Python Developer - Web Developer

- Trained models on textual data to quickly identify similar network failures previously encountered. - Designed L2 features for 5G base stations, increasing product throughput by a factor of eight.

- Mapped CO2 emissions on Earth using NASA's OCO-2 satellite data, detecting local carbon peaks.

- Developed innovative analytical solutions using Python/C++ based on clustering, optimization, and probabilistic models, increasing 5G throughput by 20%.

- Analyzed the CrunchBase dataset to predict startup success by defining key metrics.

- Assisted a project manager in selecting a cost-effective solution to reduce 3G/4G network congestion, collaborating with vendors and testing solutions in the lab.