Anshul M.

I’m a Python AI/ML engineer focused on turning raw data into something useful: insights, models, dashboards, APIs, and automated data pipelines that can run reliably every day. I’m Top Rated Plus on Upwork with a 95% Job Success Score, and most of my work sits at the intersection of Python, machine learning, AI analytics, web scraping, and data engineering. I can help if you have data but are not yet getting enough value from it, or if you are still doing too much of the process manually. What I work on • AI analytics for customer reviews, feedback, support data, survey responses, and other unstructured text • Sentiment analysis, classification, topic extraction, LLM summaries, and AI-assisted reporting • Machine learning models for prediction, classification, forecasting, scoring, and pattern detection • Web scraping and crawling for websites, marketplaces, directories, search results, competitor data, and other public sources • ETL and data pipelines that clean, transform, validate, and move data into PostgreSQL, BigQuery, Snowflake, or other databases • Dashboards and internal tools using Power BI, Streamlit, Plotly, and Python • Python APIs and backend services using FastAPI • Trading scripts, data analysis, technical indicators, and strategy backtesting in Python For me, the useful part is not just building a model. It is making sure the full workflow actually works. For example, a review-analysis project might involve collecting thousands of reviews, cleaning the data, removing duplicates, detecting sentiment and themes, generating summaries, storing the results, and presenting them in a dashboard that can be refreshed automatically. A prediction model is similar. The model itself is only one part. We also need to think about the quality of the input data, how often it changes, how the model will be evaluated, where predictions will be stored, and how the output will be used by your team or product. Recent work One long-term project involved building and continuously improving an AI-powered Amazon review analyser. The system processes customer reviews, identifies patterns and sentiment, and turns large volumes of feedback into something the business can actually act on. I’ve also worked on AI sales chatbots, where the AI had to understand user intent, respond using the right business context, and connect with the wider sales workflow. Another project involved building a Google web-scraper add-on for a digital-footprint product. The goal was to collect structured information from search results and websites and pass clean data into the rest of the platform. Web scraping is an area where I pay a lot of attention to reliability. A scraper that works once is easy. A scraper that runs every day is a different problem. I normally think about pagination, retries, rate limits, duplicate records, browser automation, failed requests, changing page structures, scheduling, logging, and how the output will be validated before it reaches the next system. The same applies to data engineering. I can build pipelines that pull data from APIs, databases, files, or websites, clean and transform it, and load it into your database or analytics stack. I also make sure there is a clear way to see when something fails rather than having a silent pipeline issue go unnoticed. My typical stack includes Python, Pandas, NumPy, Scikit-learn, PyTorch, FastAPI, PostgreSQL, BigQuery, Snowflake, Streamlit, Plotly, BeautifulSoup, Scrapy, Selenium, Playwright, OpenAI, Claude, and cloud services such as AWS and GCP. How I usually approach a project I start with the data. What do you already have? Is it clean enough to use? Is there enough of it? What result are you trying to produce, and how will we know whether the output is good enough? Once that is clear, I build a first version using real data so we can see actual numbers rather than discussing the solution only in theory. From there, I iterate toward the target, improve accuracy or reliability, automate the workflow, and deploy it with documentation. I also try to be realistic about where machine learning is needed and where a simpler solution will work better. Not every problem needs a complex model, and I would rather recommend something maintainable than overengineer the project. As part of HestaBit, I can also bring in full-stack, mobile, cloud, or DevOps support if the Python or ML component needs to become part of a larger application. So if you need a model inside a web platform, a scraper feeding a dashboard, an AI analytics engine inside a SaaS product, or a data pipeline connected to an existing backend, we can handle the surrounding engineering as well. Send me a sample of your data, the website you want to scrape, or the outcome you are trying to achieve. I can tell you what is realistic, what approach makes sense, and where the main technical risks are before we start building.