Limit Break, a blockchain game company. Limit Break is bringing the free-to-play gaming experience to Web3 and beyond. They were founded by international pioneers in the mobile gaming business.

• Built a pipeline to analyse ad campaigns, clickstream, and customer surveys data, to identify an increasing demand and launch a strategic product line. Built out a Druid dashboard for the visualtization and Prometheus for the metrics analysis.
• Built out an internal tool to categorize and automate NFT giveaways based on pre-defined templates with various weightages such as Bitcoin holdings, Etherium stockpiles, Dune metrics and previous track records of participants.
• Designed and executed A/B tests, performed rigorous statistical analysis for collaborators such as Adjust and Apsflyer based on latency, update intervals and real-time data throughput. Led to 13% MoM increase in the conversion rate.

BlueOcean is a market leader in assisting decision-making with its AI-powered insights that go beyond traditional brand trackers and point solutions from Share of Voice investments to messaging strategies to outmaneuvering the competition.

• Helped categorize and sort data as well as metrics from various sources and automated all the manual driven processes for data analysis such.
• Built out Quicksight dashboards and set up AWS stack pipeline and Tableau for visualization and reports.
• Improved data cleaning and preprocessing of data and optimized the data storage for faster query executions. Automated the sorting, classification and labeling of relevant data at cron schedules.
• Worked on streaming for real-time data (approx. 230ms latency) processing pipeline, alongside Apache Airflow for efficient management of batch jobs along with job orchestration to derive actionable insights from the massive and complex datasets.

Fable is an AI app for entertainment discovery and communities. It is on a mission to help build safe communities as cohorts of colleagues come together to discover, discuss, and organize books and shows as a means to learning and mental wellbeing.

• Worked on data processing and storage from multiple sources. Converted the unstructured data into a structured and labelled format.
• Worked on a chatbot powered by a fine-tuned llama2 Generative AI model that was trained using custom data that would suggest users a good read based on their likings, requirements and previous preferences. The bot could also give a brief summary of the book or show including a no-spoiler version to help keep the excitement of the user up. The model had an average accuracy of over 92%.
• Set up RAG implementation so the model can retrieve relatively new data such as new books and shows from the database and mitigating any need for further fine-tuning based on the new data.
• The introduction of Book Families and Book Series facilitated deeper exploration of thematic connections, while refinement of genre categorization enriched the catalogs search and recommendation capabilities. The ongoing maintenance processes ensured that the catalog remained current and relevant in a dynamic literary landscape.

Grin is the world's first creator management platform that turns brands into household names and is listed #1 across all top review sites including Capterra, G2 Crowd and Influencer Marketing Hub and collabrated with some of the world's fastest-growing brands including SKIMS, Warby Parker, Allbirds, Mejuri, and MVMT.

• Helped transition the data processing workflows from Xplenty to Apache Airflow. Airflow's modular and scalable architecture offered us the flexibility to design, schedule, and also monitor complex data pipelines efficiently. The migration was executed seamlessly, ensuring that no data was lost during the transition.
• Recognized the opportunity to optimize the data querying and transformation processes. Thus decided to replace BigQuery scheduled queries with DBT cloud, a transformation tool that provided enhanced control and flexibility. This streamlined data transformations and significantly improved processing efficiency.
• This helped reduce processing times and improve query performance. Jobs that previously took days with Xplenty were completed faster with Airflow, reducing execution by almost 7.5x time. It further enhanced Scalability as we were able to scale all our data processing capabilities effortlessly, allowing growing data volumes and diverse sources. Transitioned from BigQuery scheduled queries to DBT resulting in more reliable data transformations.