Back in middle school, I began my coding journey with a C program, but the catch was, I coded on paper. I started coding in notebooks even before I owned a computer. I completed my first notable project—an IoT device to make life easier for my farmer parents, automating water taps based on soil moisture data. That’s when I realised the real-world impact of the software I could conjure up. 

Fast forward to engineering college in Kerala, and I hit a bump in my excitement and love for this field of study. My struggles since childhood, rooted in my neurodivergence and the othering of my caste, race, and gender identity, became more apparent. This marked the first time I actually started understanding my marginalisation. I felt people discriminated against me and held prejudices in various pursuits at college, prompting me to channel my frustrations into art. My canvas became an Instagram page; hence, @huesonmycanvas was born, where I shared not just my struggles but broader societal issues that demanded attention. 

As I grew more vocal about socio-political issues and reclaimed my space as a marginalised individual on the internet, I found myself drowning in relentless online attacks and hate campaigns. Reporting abuse was challenging due to Instagram’s limitations in detecting Hinglish slurs. I observed that many artists and activists from oppressed communities were facing similar issues. The vitriol took a huge toll on my mental health, but instead of succumbing to the hate, I decided to flip the narrative. 

Thus, I birthed ShhorAI – my brainchild designed to combat online hate speech against marginalised communities, with a special focus on Indian vernacular and mixed languages. It’s not just a tool; it’s a shield against the toxicity that plagues our online spaces. 

One of the major problems with current AI solutions in multilingual spaces is the lack of effectiveness in addressing mixed-code scenarios. Gaps and lags in existing systems hinder their ability to accurately detect and handle hate speech that occurs in code-mixed forms. This limitation results in a significant impact on users’ ability to report abuse and bullying effectively. 

Building a dataset on Hinglish hate speech from scratch was a colossal task, one I couldn’t tackle alone. Seeking help on my Instagram page, amazing individuals from diverse backgrounds volunteered to tag thousands of comments, breathing life into the AI model. The hate I faced became the weapon to dismantle the very systemic oppression that fuelled it. 

ShhorAI doesn’t just detect hate in the conventional sense. It’s a multi-faceted guardian – identifying hate in code-mixed forms, transcending from text to images and videos, and even seeing through obfuscation. It’s a first-of-its-kind AI-powered first-aid kit for online hate victims, offering legal guidance, technical help, and mental health resources. 

Sitting in my room in Tripura, facing the unique challenges of being an Adivasi from a remote village, neurodivergent, and questioning gender and sexuality, my reflections on hardships bring a sense of clarity. The peaks of success attained through sheer perseverance, the profound lows, and the moments of autistic burnout – they all fall into place now. For much of my life, I’ve been a lone wolf, navigating an abyss with no map or footsteps to follow. However, this isn’t the conclusion of my narrative. Our journeys deserve the warmth of community, a shared space to flourish. Hence, by sharing my story, I aim to be a connection point, not an endpoint. I hope that those facing similar struggles may find resonance, comfort, and perhaps glean some guidance from the narrative I unfold. 

May the resistance live on through our stories, our art, and our work. May the counter-narrative keep growing stronger.