Raktim Bora

Background

About Raktim Bora

Welcome to my profile hope you are having a good day! I have been around writing code professionally for 12+ years now and the term that describes me best is "tinkerer". While formal education in Computer Science has given me a solid foundation most of the technologies I work on today have been self-taught and without a university class. I started in 2011 as a Data Warehouse developer (SQL routines were the rage back then!) and then stumbled upon ML during my M.Sc. I wrote my Master thesis in NLP transitioning into a Data Scientist role. My initial days as a Data Scientist were what I believe will be called in today's terms - Data Engineering (i was mainly doing ETL with PySpark Kafka and HBase). Then around 2018 Deep Learning took the world by storm and I couldn't resist exploring - thus starting my journey into Computer Vision. My first works were training classifiers - with no knowledge of classical CV - which I learnt much later. While training models from scratch and experimenting with optimisers and loss functions were fun initially (and still is!) the enormous progress in AI meant a lot of SOTA models are now open source and I found myself more and more fine-tuning rather than training from scratch. I also discovered that the biggest hurdle was taking these models to production optimising for speed and cost was a real problem and I have been spending more and more time focusing on this topic recently. I always love to take on interesting new challenges so if you have a project to discuss please reach out! Thank you. Github 👉 https://github.com/borarak Blog 👉 https://medium.com/@reachraktim Tech stack 👉 Python C++ CUDA Swift TensorFlow PyTorch gStreamer NVIDIA DeepStream Docker Google Cloud AWS MlOps