About Me
Hi! Welcome to my portfolio
I have close to 10 years of experience as a backend engineer working with a variety of technology stacks and domains. To me, Software Engineering is the application of Math, logic, and creative problem-solving. Being curious is my strength, I would say.
I often spend time on weekends trying to understand how systems work under the hood. This involves some experimenting and reverse engineering. This helps me develop a more holistic, finer outlook on computer science and eventually to develop better solutions at work.
Current Focus: AI Safety & Interpretability
I have recently pivoted my focus toward AI safety research and am open to exploring that space. I have been rigorously upskilling with a focus on mechanistic interpretability, reinforcement learning, and model internals.
Key Technical Milestones:
- Curriculum: Completed ARENA (AI Alignment Research Engineer Accelerator) modules
- Research: Successfully replicated key research papers from Anthropic's interpretability team
- Open Source: Contributed to safety-focused tools and open-source model development
- Foundations: Deepened expertise in advanced mathematics (Linear Algebra, Analytic Geometry, and Probability) required for high-level alignment work
- Core Stack: PyTorch, Einops, TransformerLens, Scikit-learn, Matrix Decomposition & Dimensionality reduction (SVD, PCA, t-SNE)
Software Engineering Background
I want to adopt a more sophisticated and comprehensive perspective on computer science. I have been delving more deeply into popular technologies that people take for granted - Databases, distributed systems, concurrency, computer graphics, machine learning, Advanced Algorithms & Data Structures, and much more.
I can never claim that I have learned enough, or that I am the best expert in town, and I guess that is okay... as long as the learning never stops. It's the progress that matters in the end, doesn't matter where the starting point is.
What I Write About
I write blog posts on topics that I find really interesting - from distributed systems and database internals to mechanistic interpretability and AI safety. Hope you find them interesting as well!