TrafficTrack

TraffiTrack: Revolutionizing Digital Drug Detection with Real-Time Intelligence

TraffiTrack—a groundbreaking system built to empower law enforcement agencies in their battle against the ever-elusive digital drug trade.

Imagine a world where advanced AI flags suspicious conversations and cryptic phrases before illegal exchanges even happen.

That’s the promise TraffiTrack holds. Created by a team of young, driven developers—none of whom are over 20 years old—TraffiTrack isn’t just software; it’s a digital watchdog designed for a safer future.

The Backstory: Why TraffiTrack Matters

Drug trafficking has taken a dark, digital turn. Traffickers leverage the anonymity of platforms like Telegram, Instagram, and WhatsApp to exchange information, using coded language, emojis, and multimedia to hide their intentions. For law enforcement, this presents a massive challenge: it’s like finding a needle in a digital haystack of messages, memes, and encrypted codes.

TraffiTrack’s mission? To make it possible for law enforcement to step in at precisely the right moment—flagging suspicious activity with surgical precision and stopping crime in its tracks.

It’s a mission that requires cutting-edge tech, real-time analysis, and relentless innovation. But here’s the twist: TraffiTrack was built by a team of college sophomores, some of whom just turned 20. And this was their first hackathon.

DRUG trafficking has gone 2.0, and the law has not kept up with criminals.

Drugs no longer hide in the shadows of streets and back alleys, they are now moving to encrypted platforms such as Telegram, Instagram and WhatsApp. Inside secret rooms, traffickers use coded language, emojis and multimedia to communicate in real time with little hope traditional tools can keep up. Law enforcement has a difficult time locating and intercepting these exchanges among the data noise and volume.

Thus, TraffiTrack comes into play. TraffiTrack enables LEA to predict drug deals before they happen by analyzing language patterns, coded slang detection and geospatial correlation. This is not only a reaction to the crime, it is also about prevention. TraffiTrack — the one monitoring your behavior online and raising a flag at even the faintest sign of something suspicious.

The Tech That Powers TraffiTrack

Watch it in action here: https://www.youtube.com/watch?v=xA9F7TYYrfA&t=29s&ab_channel=AsmiParikh

Our team, all second-year students under 20, had minimal exposure to the level of tech TraffiTrack demanded. But we had a vision, and we learned fast—picking up new tools, exploring uncharted tech, and committing countless hours to make TraffiTrack a reality. Every day was a new challenge, but we were relentless, determined to see TraffiTrack come to life.

Some of the tech highlights we managed to pull off include:

1. Secure Access Control with Auth0: Security is number one on the list. We added layers of authentication and authorization to TraffiTrack and with Auth0 we could trust that it is now secure and compliant for law enforcement able to use seamlessly. For a team that was new to this scale of user security, it was a steep curve—but one that taught us the significance of building responsibly and efficiently.

2. Creating Real-Time & Interactive Dashboards with Streamlit: Instead, Streamlit was our friend to get the user-based and interactive interface we wanted. The TraffiTrack dashboard not only displays but also interactively guides the users through real time alerts, location based maps and all the patterns that can be monitored. Being streamlit noobs, we were floored by its capability, and after hours of tinkering and crafting we made a dashboard that seamlessly allows for crime tracking.

3. NLP Superpowers Using LLaMA Language Models: From slang to emojis, we used LLaMA to dine on crypto drug talk. The NLP system developed by TraffiTrack is able to detect potential threats even when they are couched in everyday talk, allowing law enforcement to detect intent even if words are hidden behind coded language. But this was no run of the mill implementation — it needed a lot of intensive training to manage nuance, context and slang correctly.

4. Geospatial and Temporal Pattern Analysis: TraffiTrack analyzes patterns not just words. We trained TraffiTrack with geolocation and temporally inferring data to identify suspicious activity based on location and time. Whether it is uncharacteristic movement in regular trouble spots or increased messaging volume, TraffiTrack has been programmed to do one better, always monitoring and learning from behaviours.

   

How we brought the TraffiTrack Experience together

We had a vision when we started TraffiTrack – to create something bigger than us. None of us had prior advance NLP and streaming data processing experience before we started. But we learned anyway, breaking things one error message at a time.

It was not just the hackathon, but a life lesson to step out of our comfort zone.

From language models to Async processing, TraffiTrack is a culmination of implementation after implementation after hundreds of hours spent coding, learning, and rethinking what can really be done.

def generate_sample_data():
    data = {
        "Drug Name": ["MDMA", "LSD", "Mephedrone", "Cocaine", "Heroin"],
        "Detected Instances": [10, 15, 7, 12, 5],
        "Flagged Users": [5, 10, 4, 7, 3],
        "IP Addresses": [3, 8, 2, 6, 2]
    }
    return pd.DataFrame(data)

def check_for_drug_content(input_text):
    drug_keywords = ["MDMA", "LSD", "Mephedrone", "Cocaine", "Heroin"]
    pattern = r'(\+?\d{1,3}[-. ]?)?\(?\d{1,4}?\)?[-. ]?\d{1,4}[-. ]?\d{1,4}[-. ]?\d{1,9}'  # Regex for phone numbers
    ip_pattern = r'(?:(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.){3}(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)'  # Regex for IP addresses

    found_drugs = [keyword for keyword in drug_keywords if keyword.lower() in input_text.lower()]
    phone_numbers = re.findall(pattern, input_text)
    ip_addresses = re.findall(ip_pattern, input_text)

    return found_drugs, phone_numbers, ip_addresses

st.title("Login")
email = st.text_input("Enter your email...")
password = st.text_input("Enter your password...",type='password')

The Struggles: Growing through Fire, Evolving through Mistakes

We have to say it as it is: this was not an easy road. NLP, data security – these were all basic building blocks of TraffiTrack that we had not yet acquired the necessary skills to solve successfully and each day was a new broadside. These are just some of the challenges we faced and overcame:

• Auth0 is new to Advanced Authentication: not going to lie, we had never seen Auth0 until now. But we are all in on security! We were knee deep in documentation, forums, and anywhere else we could find answers. We got a secure, scalable and compliant authenticaton layer for TraffiTrack after painting it out on the walls of our rooms repeatedly over few random days

• Aligning Multimodal Data to Real Time: It did not take long before the team realized how challenging it was to integrate text, image and geospatial data. Further, these three data streams had to be coordinated which did not come without its challenges; however, given our system took advantage of concurrent processing with extensive use of TensorFlow Operations (Ops), we were able to maximize integration. We struggled through the steep learning curve, and eventually made forsythe a system that works like grease and can be expanded as social media finds new outlets to become popular.

• Learning a Second Language: Our coded language — training a model to identify drug slang: At first, we encountered a lot of false positives — suspicious-sounding phrases that were merely coincidences. Using weeks of finetuning plus in-the-field testing, TraffiTrack's language model is now capable of separating normal chatter from genuine drug-speak with surprising precision.

A New Era of Digital Crime-Fighting

And this is the beauty of TraffiTrack! Designed for scale, TraffiTrack can scale up to encompass any new social media platforms that come onto the scene, providing law enforcement with a constantly evolving advantage over digital drug networks。 It is more than crime response — it is about being proactive.

TraffiTrack is a new generation of digital, real world impact tools using psycholinguistic profiling, geospatial analysis and live monitoring. TraffiTrack will be a key tool in the digital war on crime and we hope that agencies can use it to target trafficking networks at a reduced cost and time.'

1. Real-Time Intelligence: TraffiTrack provides law enforcement with real-time alerts, ensuring immediate action where it’s needed.

2. Advanced Scalability: Our concurrent processing model means TraffiTrack can handle massive volumes of data, ready to grow as new social platforms enter the scene.

3. Intuitive, Interactive UI: Thanks to Streamlit, TraffiTrack’s dashboard is built for ease and engagement, giving users a seamless experience that enhances situational awareness and allows them to drill down into data with precision.

What’s Next for TraffiTrack?

For us, TraffiTrack is only a starting point. We are already thinking about new features, better algorithms and further ways to make TraffiTrack smarter. We look forward to further developing this project, and ultimately transforming TraffiTrack from a hackathon prototype into a proposed tool for law enforcement agencies around the world.

Conclusion: Built by Heart and Driven By Passion

TraffiTrack is not just a piece of code, it is our passion project, our dream and proof that the impossible for a group of young women can become possible only if you put some purpose behind it. In our first hackathon, as students we pushed the limits of our capabilities and where we thought we could take technology to make a difference in the world.