Just wrapped an automation for a client racing against a January 31st contract deadline. They needed to backup thousands of course and program records with all associated PDFs before losing access to the platform - what would've taken their team weeks of manual clicking, downloading, and organizing. I built a Python script using requests and Selenium that handles the entire workflow: logs into the platform, filters by historical and active status across all subjects, extracts course and program information, downloads all available PDFs, converts HTML pages to PDF when needed, renames everything with clear file naming conventions, and organizes it all into structured folders. The client went from facing a data loss crisis to having a complete backup ready before their deadline. Anyone else deal with tight deadlines recently?

Here's how I scraped 30,000+ encrypted videos from a protected CDN for my client. When you're planning to download photos or videos, you may not immediately see them in the Network tab. To make it easier, change the filter from "All" to "Media". This will show you the requests you're looking for. Once you see the requests being made, you will most likely encounter one of the following: 1. Video stream chunks (.ts segments), not single .mp4 files In this case, you have to bypass the CDN, get each chunk, and stitch the video together to make the .mp4 file. Many CDNs use AES-128 encryption to protect content. You need to: - Extract the encryption key URI from the .m3u8 playlist - Download the decryption key - Grab the initialization vector (IV) Without all three, you can't decrypt anything. The last segment has PKCS7 padding that needs to be stripped. Skip this step and your final video will be corrupted. Concatenate all decrypted segments into a single .ts file, then use ffmpeg to convert to .mp4: ffmpeg -i combined.ts -c copy -bsf:a aac_adtstoasc output.mp4 The -bsf:a aac_adtstoasc flag is critical for proper audio encoding. 2. Full MP4/image files are available (the easier case) Sometimes you'll find a direct link to the full .mp4, .jpg, or .png file. In this case: - Right-click the request in the Media tab → Copy as cURL - Paste the contents into a cURL to Python converter or have AI implement it for you - Even when videos aren't encrypted, CDNs still protect them. You need to mimic a real browser request to succeed. Replace requests with curl_cffi and ensure you can make a request without getting blocked. Watch out for: - Expiring URLs with time-based tokens (you'll need to regenerate them) - Rate limiting (add delays between requests or rotate IPs) - Dynamic URLs that change per session

Built a court records automation for a client who needed to search 1,700+ terms across the New Jersey Courts website. Doing this manually would've consumed hundreds of hours of their team's time - opening the site, entering each search term, collecting results, copying data, organizing everything into a usable format. The automation handles the entire workflow: systematically searches all 1,700 terms, extracts case information from each result, and consolidates everything into a clean CSV file ready for analysis. What would take weeks of manual courthouse research now runs overnight while the client focuses on actually working their cases instead of gathering data. Full automation demo: https://www.loom.com/share/cc5d22635e924464989661dded583f8d What repetitive research tasks are eating up your team's time right now?

Just built and delivered an automation for a client who had thousands of probate cases that needed to be augmented with court records from a specific website. Manually looking up each name, downloading the relevant PDFs, and adding links to the spreadsheet would've taken weeks of tedious work - so I created a Python script that handles the entire process automatically. The system opens a browser, reads the CSV file, searches each person's name on the court website, collects all associated documents (petitions, notices, orders for service, etc.), downloads the PDFs, and adds everything back to an updated CSV file. What would have been thousands of manual searches and downloads now runs completely hands-off while the client focuses on actual casework instead of data entry. Full automation walkthrough: https://www.loom.com/share/49763d1dbcac4906bfa0da70d28cfdce What's an automation you wish your business had?

Just wrapped up a custom property management platform for a client that handles everything from property scraping to agent-client collaboration. The system pulls apartment listings from multiple sources, displays them on an interactive map with custom radius filtering, and lets users filter by price, bedrooms, bathrooms, and square footage in real-time. The coolest part is the agent workflow - agents can create curated property lists for their clients, clients review and select their favorites, and agents instantly see what their clients liked without any back-and-forth emails or calls. Built features like magic link authentication for instant access, favorites management, admin dashboards showing all users and activity metrics, and dynamic filtering at both the portfolio and individual property level. When you need software that fits your exact business workflow instead of forcing you into generic SaaS limitations, custom development makes all the difference. Full platform walkthrough: https://www.loom.com/share/48da24b65b65403897bb51401e3ebb42 What custom software would transform your business operations if you could build it exactly how you wanted?