AI-Augmented Data Analyst & Full-Stack Developer | Python, JS, ML

Python is my go-to language for data analysis, machine learning, and automation. My day-to-day work happens in Jupyter Notebooks — exploratory data analysis, data cleaning with pandas and NumPy, building and evaluating ML models with scikit-learn, and working with NLP libraries like Transformers and sentence-transformers. On the ML side specifically, I've fine-tuned two custom embedding models — a Nomic BERT and a BGE-M3 — trained on over 42,000 triplets, and built the full retrieval pipeline around them including FAISS vector search, BM25, and reranking logic. I also use Python for scripting and automation — data pipelines, API calls, Google Sheets integration, and ETL work. I do have a Django background from earlier in my career, but that's been several years and I wouldn't claim it as current. My Python strength today is firmly on the data and ML side.

My AI experience spans three layers: using AI tools in my daily workflow, building systems that incorporate AI and ML, and working as a human data professional in an AI pipeline. On the tools side, I use Claude, ChatGPT, and GitHub Copilot regularly — for drafting, code assistance, data analysis acceleration, and problem-solving. I treat them as force multipliers, not replacements for critical thinking. On the development side, I've fine-tuned two ML models from scratch — one based on Nomic BERT and another on BGE-M3 — trained on a custom dataset of over 42,000 triplets for semantic search and retrieval. I built the full pipeline around them: embedding generation, vector search using FAISS/HNSW, ranking algorithms, and a retrieval system with 18 processing stages including BM25, k-nearest neighbors, and reranking logic. That system is live at cap-teyyko.live. On the data and human-in-the-loop side, I worked at micro1 as a Human Data Manager, where my role was directly inside an AI training pipeline — evaluating model outputs, improving data quality, and providing the human judgment layer that makes AI systems more accurate and reliable. So I understand AI from the inside out: as a user, as a builder, and as someone who has worked within the human feedback loop that trains these systems.

I have extensive experience using Microsoft Excel for data analysis, reporting, and tool development. As a Data Analyst/Developer at Cheetah, I utilized Excel to streamline analysis processes, create automated reports, and develop tools for labor planning, performance reviews, and inventory tracking. My skills include advanced functions, VBA programming, and custom tool creation to enhance operational efficiency.

I use Google Sheets well beyond basic spreadsheets — I build operational data systems in it. At my most intensive, I was managing a 600-participant event where Sheets served as the live backend synced to a Flutter mobile app via a Google Service Account and the Sheets API v4. The app read and wrote participant data bidirectionally in real time, while Sheets itself contained structured dashboards with committee-level analytics, dynamic summary tables, and COUNTIFS-driven statistics that updated automatically. On the formula side, I work comfortably with VLOOKUP, INDEX/MATCH, ARRAYFORMULA, QUERY, IMPORTRANGE, conditional formatting rules, and data validation with dependent dropdowns. I've also done CSV reconciliation work — matching two datasets of 500+ names using fuzzy logic principles translated into spreadsheet logic, flagging mismatches, and resolving duplicates. I've built reporting templates for non-technical users where all they do is paste raw data and the summary auto-populates. I understand Sheets both as an end-user tool and as a programmable data layer.

I have extensive experience with SQL and database management, specializing in designing, optimizing, and maintaining efficient data systems. At Cheetah, I utilized SQL to extract, analyze, and visualize data for business development, creating tools like a heatmap for warehouse locations and dashboards for labor planning and performance reviews. My work also included developing automated reporting processes for daily, weekly, and monthly insights, ensuring accurate and actionable data for decision-making. I’ve worked with databases such as Microsoft SQL Server, Amazon Redshift, and Google Sheets, integrating them seamlessly with Python, FastAPI, and other technologies for comprehensive solutions.

I have extensive experience using FastAPI for backend development, focusing on building high-performance, scalable web applications. At Cheetah, I developed the Safety App using FastAPI to create a robust backend for incident reporting, safety dashboards, and real-time notifications. This project involved seamless integration with machine learning models and APIs for enhanced functionality. I also leveraged FastAPI in developing the Loading Assistant, which optimized truck loading based on volume, routes, and other parameters. My expertise includes crafting RESTful APIs, integrating them with frontend technologies like Vue.js, and ensuring backend systems are secure, efficient, and maintainable.

JavaScript is one of my core languages and I use it across the full stack. On the frontend, I work with React — building component-based UIs, managing state, handling API calls, and structuring projects for maintainability. I've also worked with Vue.js in previous projects. On the backend, my primary JS runtime is Node.js with Fastify, which I've used to build production REST APIs — including a live semantic search application with real-time features via Socket.IO. I'm comfortable with async/await patterns, middleware design, route structuring, and integrating with databases like PostgreSQL and SQLite. I also have Electron experience for desktop app development, and I've written vanilla JS for scripting, DOM manipulation, and browser-side logic when frameworks aren't needed. So across the stack — React on the front, Fastify/Node on the back, with real deployed projects — JavaScript is one of the languages I'm most confident in.